SabrTrader Documentation

Trade Performance Analytics and the AI Coach

Trade Performance is SabrTrader's post-trade analytics workspace. Opened from the Control Center's New menu, it reads your executed trade history, reconstructs it into round-turn trades, and presents it as statistics, charts (equity curve, drawdown, distribution, period aggregation, Monte Carlo) and an AI Coach review generated by the Explain My Results button. It serves discretionary traders reviewing a session, algo traders validating live results from AlgoStudio Pro or the Auto Strategy Builder, and funded-account traders checking risk compliance against daily loss limits.

Overview: What the Trade Performance Window Is

Trade Performance is the SabrTrader window that performs post-trade analytics: it takes the fills recorded in your trade history, groups them into completed trades, and converts them into performance statistics, charts and an AI-written review. It does not place orders and it does not analyse open market conditions. It answers one question only: what actually happened in the trades you already took?

The window is opened from New > Trade Performance in the Control Center. It opens as its own top-level window, which means it can be moved to a second monitor, resized independently of your charts, and left open on a review monitor without covering trading tools.

Who the window is for

• Discretionary traders — end-of-day and end-of-week review of manual trades: win rate, average win versus average loss, drawdown depth, and which instruments or hours produced the results.
• Algo and strategy traders — measuring the live results of a strategy built in AlgoStudio Pro or generated by the Auto Strategy Builder, rather than relying only on backtest output.
• Funded-account and prop-firm traders — comparing realised drawdown and worst-day loss against the account's daily loss limit and maximum drawdown rule.

What the window contains

Area	Purpose
Date range selector	Defines the analysis window: Today, trailing 12 months, prior calendar years, or a custom start/end date.
Account filter	Restricts statistics to a single trading account (sim, live, funded, or a strategy-dedicated account).
Instrument / asset filter	Restricts statistics to one symbol so results can be attributed per market.
Analysis tabs	Trades, Lots, Equity Curve, Drawdown, Distribution, Periods, Monte Carlo.
AI Coach panel	Generates a written review of the currently filtered dataset via Explain My Results.

Key principle: every number, chart and AI sentence in the window describes only the dataset selected by the filters. The filters are therefore the most important control in the window — they define the sample, and the sample defines whether any conclusion is valid.

Opening the Window and Placing It on a Monitor

Trade Performance is a launchable window, not a chart overlay. It follows the same modular window model as the rest of the platform described in platform setup and workspaces.

1. Open the SabrTrader Control Center (the main platform window).
2. Click New.
3. Select Trade Performance.
4. The window opens on whichever monitor the platform last used for a detached window. If it appears on a different screen than expected, drag it by its title bar to the monitor you want.
5. Resize the window by dragging its edges. Widening the window widens the filter row and the tab bar, so the date range, account and instrument fields are fully readable and the analysis charts get usable horizontal space.

Recommended placement

• Dedicated review monitor. Keep Trade Performance on a secondary display or a separate virtual desktop so it never covers your DOM, footprint or charts during the session.
• Maximised for review sessions. The Equity Curve, Drawdown and Monte Carlo tabs are shape-reading tools; a small window compresses the horizontal axis and hides the shape you are trying to read.
• Closed during execution. Watching a live equity curve mid-session encourages P&L-driven decisions. Open it after the session, not during it.

Filter Reference: Date Range, Account and Instrument

The filter row at the top of the window defines the dataset. Nothing below the filter row means anything without knowing how the filters are set.

Date range selector

The date range selector chooses the time window whose closed trades are included. It offers presets plus a custom range:

Option	Includes	Typical use
Today	Trades closed during the current session date.	End-of-day review; the AI Coach example in the demonstration used Today and found 27 trades.
Trailing 12 months	The last twelve months of history, rolling.	Annualised statistics, long-run profit factor and maximum drawdown.
Prior calendar year(s)	A complete past year, e.g. last year, the year before.	Year-over-year comparison; checking whether an edge persisted.
Custom range	Any explicit start and end date you enter.	Isolating a strategy's live run window, a specific week, or the period after a rule change.

The historical reach of the report is limited by how much trade history exists on the machine and account. If the platform was not running or not connected on a given day, no fills were recorded for that day, and the report will show a gap rather than a zero.

Account filter

The Account filter restricts the report to a single trading account. This is account segmentation: keeping simulated, live and funded results in separate reports so their statistics are not blended. Blending a sim account (no slippage penalty, no emotion, sometimes oversized) with a live account produces statistics that describe neither.

Instrument / asset filter

The Instrument (asset) filter restricts the report to one symbol, for example MES or MNQ. This is instrument-level attribution: it reveals that a flat overall equity curve can be a profitable symbol plus a loss-making one, hidden by aggregation.

Blank-filter behaviour

Leaving a filter blank includes everything. A blank Account field means all accounts; a blank Instrument field means all symbols. This is convenient for a total picture and dangerous for interpretation, because it silently merges sim, live and demo trades into one report.

Filter stacking

Filters combine with AND logic. Each additional filter narrows the sample:

• Today + blank account + blank instrument = every trade you took today anywhere.
• Today + live account + blank instrument = today's live trading only.
• Last 12 months + live account + MES = one year of MES results on the live account.
• Custom 3 Feb – 28 Feb + strategy account + MNQ = one automated strategy's February run.

Every narrowing step reduces sample size. A tightly filtered report is more specific but statistically weaker. Always note the trade count that the report shows before drawing a conclusion from it.

Tab Reference: Every Analysis View and How to Read It

The tab bar switches between seven views of the same filtered dataset. Each tab answers a different question.

Trades tab

The Trades tab lists each completed trade as one row: instrument, direction, entry, exit, quantity, timestamps and net profit or loss. It is the product of round-turn trade reconstruction — individual fills, including partial scale-ins and scale-outs, are grouped so that one intention becomes one measurable trade.

• Healthy pattern: losses cluster in a narrow band (consistent stop discipline); wins vary but the largest loss is not larger than the largest win.
• Warning sign: one or two losses several times larger than the rest — a stop was widened, averaged into, or not honoured.

Lots tab

The Lots tab breaks the same activity down to the contract (lot) level rather than the trade level. Where the Trades tab says "one trade, 3 contracts, +$120", the Lots tab shows the per-contract execution detail. Use it to:

• Audit scaled entries and exits fill by fill.
• Compute per-contract efficiency (net profit ÷ contracts traded), which is the fairest comparison when your position size changed during the period.
• Verify total volume executed for commission reconciliation.

Do not compare Trades counts with Lots counts. Ten trades of three contracts is ten trades and thirty lots. Confusing the two inflates or deflates every per-trade statistic.

Equity Curve tab

The Equity Curve tab plots cumulative realised profit and loss across the filtered period or trade sequence.

• Healthy pattern: a stair-step rise — small pullbacks, higher highs in equity, no single vertical jump carrying the whole result.
• Warning sign: a choppy, flat curve with heavy oscillation (edge is inside the noise), or a curve where one enormous trade produces all the gain (result is an outlier, not a process).

Drawdown tab

The Drawdown tab plots the decline of equity from its running peak. Read two dimensions: depth (how much was given back) and duration (how long it took to recover).

• Healthy pattern: maximum drawdown is small relative to your average winning day and well inside your account's loss limits; recovery takes a few trades or a day, not weeks.
• Warning sign: maximum drawdown approaches your prop firm's trailing drawdown or your own daily loss limit; or drawdowns get progressively deeper over time, which suggests either degradation or increasing size.

Distribution tab

The Distribution tab is a histogram of trade outcomes: how many trades landed in each profit and loss bucket.

• Healthy pattern: losses tightly grouped around one bucket (your planned risk); wins spread to the right; few or no losses beyond the planned risk bucket.
• Warning sign: a long left tail (occasional outsized losses), or a distribution where removing the two best trades turns the period negative.

Periods tab

The Periods tab aggregates results by calendar period — daily, weekly and monthly. It measures consistency rather than total profit.

• Healthy pattern: the majority of periods positive, the worst period comparable in magnitude to the best.
• Warning sign: one huge month carrying a flat year; or one recurring weak day of the week or weak session hour, which points to a time-of-day performance problem you can fix by simply not trading that window.

Monte Carlo tab

The Monte Carlo tab randomly reshuffles and resamples your historical trade results many times to generate a range of alternative equity paths. Its purpose is not prediction; it is to show how much of your actual result was sequence luck.

• Read the spread of outcomes: how good and how bad the same edge could plausibly have looked.
• Read the worst simulated drawdown: this is a more honest sizing input than the drawdown you happened to experience.
• Read the implied risk of ruin: the share of simulated paths that lose a defined portion of equity at your current sizing.

• Healthy pattern: nearly all simulated paths end positive and the worst simulated drawdown still fits inside your account limits.
• Warning sign: a large fraction of simulated paths end negative, or the worst simulated drawdown would breach your funded-account rule. Both mean the position size is too large for the measured edge, even though the actual history looks acceptable.

The AI Coach: Explain My Results

The AI Coach is a panel inside the Trade Performance window that generates a written, structured review of the trades currently selected by your filters. It is triggered by the Explain My Results button.

How it works

1. Set the date range, account and instrument filters to define exactly the dataset you want reviewed.
2. Click Explain My Results.
3. The coach reads the filtered trade set — trade count, wins, losses, sizes, sequence, timing — and returns a written analysis.

The filters are the prompt. The coach evaluates only what the filters selected. Reviewing "Today, live account, all instruments" and reviewing "Today, all accounts" will produce different coaching, because they are different datasets.

What the output contains

The generated analysis is organised around a short set of questions:

Output block	Content
Dataset summary	How many trades were analysed and over what window. In the demonstration the coach reported 27 trades for the Today filter.
What is strong	Positive, evidence-based observations — for example a solid win rate combined with a favourable win/loss ratio, and a tightly controlled losing side.
What is weak	Statistical weaknesses: loss size dispersion, poor performance in a particular hour or symbol, evidence of overtrading.
What stands out	Anomalies: outlier trades, unusual clustering, sudden size changes.
Concrete items to work on	Specific, actionable suggestions — adjustments, modifications and risk management controls such as trade caps or loss limits.
Limitations	Things the coach explicitly cannot coach around, most commonly data gaps caused by the platform being started and stopped during the day, which leave incomplete history.

Why the output is useful

The coach provides objective performance feedback: a data-driven read that is not shaped by recency bias, by how the last trade felt, or by the desire to justify a decision. It reads the whole filtered sample with equal weight.

The coach works on winning periods and losing periods. Winning periods give confirmation — useful, but low information. Losing periods are where the value is highest, because the coach names the mechanism behind the loss (loss size, frequency, timing, instrument) instead of leaving you with a bad feeling and no rule change.

Requirements and boundaries

• The AI Coach uses a configured large language model provider. If no provider or API key is set up, configure one first — see AI setup: connecting an LLM provider.
• The coach analyses closed trades only. Open positions carry unrealized P&L and are not part of the review.
• The coach is a review tool, not an execution tool. It does not modify strategies, alerts or orders. To turn an insight into automation, implement it in AlgoStudio Pro or as a condition in the alert builder.
• Coaching quality scales with sample size. A four-trade filter produces a thin, generic review; a 100-trade filter produces specific, testable observations.

Trading Concepts Behind the Numbers

Every statistic in Trade Performance is defined below. Use these definitions consistently; most review errors are definition errors.

Core outcome statistics

Metric	Definition	How to read it
Win rate	Closed profitable trades ÷ total closed trades, as a percentage.	Meaningless alone. A 70% win rate with losses three times the size of wins loses money.
Win/loss ratio	Average winning trade size ÷ average losing trade size.	Above 1 means your average winner is bigger than your average loser.
Profit factor	Gross profit ÷ gross loss.	Above 1.0 is net positive. 1.0–1.2 is fragile; 1.5+ over a large sample is robust.
Expectancy	Average expected profit per trade: (win rate × average win) − (loss rate × average loss).	The single most useful number. Positive expectancy × enough trades = profit.
R-multiple	Trade result expressed as a multiple of the risk taken on that trade. Risking $100 and making $250 is +2.5R.	Normalises results across different position sizes so trades are comparable.

Risk statistics

• Drawdown is the peak-to-trough decline in equity. It is the cost of holding the edge.
• Maximum drawdown is the largest such decline in the selected period. It is the primary input to position sizing and the primary constraint in funded-account rules.
• Risk of ruin is the statistical probability of losing a defined portion of account equity given your current edge and sizing. It rises steeply with size and falls with expectancy.
• Daily loss limit is a hard stop on losses for one session — self-imposed, or mandated by a prop firm. Compare it to the worst day in the Periods tab and to the deepest drawdown in the Drawdown tab.
• Risk management controls are the rule set that caps damage: daily loss limit, maximum contracts, maximum trades per session, mandatory stop on every entry.

Distribution and simulation

• Trade distribution is the histogram of outcomes. Its shape tells you whether results come from a repeatable process (tight loss cluster) or from luck (a few giant wins).
• Distribution skew describes which tail is longer. A long right tail with a short left tail is the desirable shape: many small controlled losses, occasional large wins.
• Monte Carlo simulation randomly reorders and resamples your trade results to produce many alternative equity paths. It separates edge from sequence luck and produces a realistic worst-case drawdown estimate.

Sample size

Sample size is the number of trades inside the filtered window, and it governs how much any statistic can be trusted. A single session of 27 trades — the demonstration's dataset — is a snapshot of behaviour, not proof of an edge. Practical thresholds:

Trades in window	What it supports
1–10	Behavioural review only (did I follow my rules?). No statistical conclusion.
10–30	Pattern spotting: recurring mistakes, one bad hour, one bad symbol.
30–100	Provisional expectancy and win/loss ratio. Enough to justify one rule change.
100+	Reasonably stable profit factor, drawdown profile and Monte Carlo output.

Accounting concepts that change the numbers

• Realized vs unrealized P&L. Trade Performance measures closed (realised) trades. An open position's mark-to-market value is unrealised and excluded, which is why the report can disagree with a live account balance.
• Commissions and fees are per-side transaction costs. On high-frequency intraday trading they can consume the entire gross edge. Always read net figures, and know whether the report you are looking at includes fees.
• Round-turn trade reconstruction groups fills into one entry-to-exit trade, so scaling in and out does not appear as many separate trades.
• Lots / contracts traded is total executed volume. Compare it with trade count to see whether a change in results came from a change in edge or simply a change in size.
• Position sizing is how many contracts you risk per trade relative to equity and stop distance. Most "strategy failures" found in this window are sizing failures.

Why a smooth equity curve happens

The combination the AI Coach praised in the demonstration — a solid win rate together with a tightly controlled losing side — is exactly the combination that produces a stair-step equity curve. Frequent wins supply the upward slope; capped losses prevent the giveback that turns a rising curve into a choppy one. Improving loss control usually smooths the curve faster than improving win rate.

Behavioural concepts

• Overtrading is taking more trades than the edge supports. Its signature in this window is a high trade count in the Trades tab with a flat or declining Equity Curve and rising commission drag.
• Trade journaling is recording context and outcome so experience becomes process. Trade Performance supplies the objective half; your notes supply the reason half.
• Data gaps are periods where the platform or connection was not active, so no fills were recorded. Gapped history distorts period aggregation and is explicitly called out by the AI Coach as something it cannot interpret.

Strategy evaluation concepts

• Algo / strategy performance review is evaluating an automated strategy's real filled results over a defined window instead of trusting its backtest.
• Backtest vs live divergence is the gap between simulated and actual performance, caused by slippage, latency, partial fills, commissions and thin liquidity. A live curve that is directionally similar but shallower than the backtest is normal; a live curve with the opposite sign is a modelling error. Rehearsing a strategy in Market Replay before live deployment narrows this gap.

Step-by-Step: End-of-Day and End-of-Week Review Routine

End-of-day review (10 minutes)

1. Open the Control Center and choose New > Trade Performance. Move the window to your review monitor and widen it.
2. Set the date range to Today.
3. Set the Account filter to the account you actually traded — your live or funded account. Do not leave it blank on day one of a review habit; blank includes sim and demo trades.
4. Leave the Instrument filter blank for the first pass so you see the total picture across symbols.
5. Open the Trades tab and confirm the trade count and that every trade you remember is present. A missing trade means a filter or a data gap.
6. Open the Equity Curve tab. Note the shape: stair-step, choppy, or one-trade-carried.
7. Open the Drawdown tab. Record the day's deepest drawdown and compare it to your daily loss limit. If you came within reach of the limit, that is the day's lesson regardless of the final P&L.
8. Open the Distribution tab. Look for losses outside your planned risk bucket and for outlier wins that carried the day.
9. Click Explain My Results in the AI Coach panel and read all output blocks, including what is weak.
10. Write down one or two actions for tomorrow — for example "no new entries after 11:30" or "maximum 8 trades". Not five actions. One or two.

End-of-week review (30 minutes)

1. Set the date range to the trailing week (custom range Monday to Friday).
2. Keep the account filter on your live account. Run a first pass with the instrument filter blank.
3. Run a per-instrument pass: set the instrument filter to each symbol you traded in turn and record net result, trade count and win rate for each. This is instrument-level attribution — it frequently shows one symbol funding another's losses.
4. Open the Periods tab in daily mode. Check day-of-week consistency: is one weekday reliably negative?
5. Open the Lots tab and compute net profit per contract. If per-trade results improved but per-contract results did not, the improvement came from size, not skill.
6. Run the Monte Carlo tab as a sizing sanity check. If the worst simulated drawdown breaches your account's drawdown rule, reduce contracts before doing anything else.
7. Click Explain My Results on the weekly dataset. Weekly coaching is more reliable than daily coaching because the sample is larger.
8. Convert the week's findings into exactly one measurable rule change, and keep it unchanged for at least 30 trades so it can be evaluated.

Month-end and year review

1. Set the date range to the trailing 12 months or a prior calendar year.
2. Filter by account to keep environments separate.
3. Read profit factor, expectancy, maximum drawdown and the Periods tab in monthly mode.
4. Compare the same metrics across consecutive years to see whether the edge persisted or decayed.

Step-by-Step: Reviewing an Automated Strategy's Live Results

Trade Performance is the tool that answers whether a strategy built in AlgoStudio Pro or produced by the Auto Strategy Builder behaves in live conditions the way its backtest promised.

Preparation: make the strategy isolatable

1. Dedicate an account or a symbol to the strategy before you deploy it. If the strategy runs on the same account and symbol as your manual trading, its results cannot be separated afterwards.
2. Record the exact start date and time of the live run, plus the version of the strategy and its parameter set.
3. Note the backtest expectations you want to test against: expected win rate, expected profit factor, expected maximum drawdown, expected trades per day.

Review procedure

1. Open New > Trade Performance.
2. Set the date range to a custom range matching the strategy's live run window exactly. Including days before deployment contaminates the sample.
3. Set the Account filter to the strategy's dedicated account, or the Instrument filter to its dedicated symbol.
4. Check the trade count on the Trades tab against the backtest's expected trade frequency. Far fewer live trades usually means missed signals, connection interruptions or session-time restrictions; far more means a condition is firing differently on live data.
5. Compare the Equity Curve shape with the backtest curve. Look for the same slope character, not identical numbers.
6. Compare maximum drawdown on the Drawdown tab with the backtest's maximum drawdown. A live drawdown deeper than anything in the backtest is a red flag.
7. Open the Distribution tab and compare average loss with the strategy's designed stop. Live losses systematically larger than the designed stop indicate slippage or fill quality problems.
8. Open the Lots tab to confirm the strategy traded the intended size and did not scale unexpectedly.
9. Run Monte Carlo to determine whether the observed drawdown is inside normal variance for the measured edge or beyond it. Inside variance means keep running and keep sizing; beyond variance means degradation, and the strategy should be reduced or halted.
10. Open the Periods tab to see whether behaviour changed over time — a strategy that worked for three weeks and then flattened has met a different market regime.
11. Click Explain My Results to get a written summary of behaviour changes and concrete weaknesses across the run window.

Interpreting backtest vs live divergence

Observation	Most likely cause	Action
Live win rate similar, average win smaller	Slippage on exits; commissions not modelled	Re-run the backtest with realistic costs; widen targets or reduce frequency
Live losses larger than designed stop	Latency, gaps, thin liquidity at stop price	Check the instrument's liquidity; consider a different session window
Far fewer live trades	Missed signals from disconnections or data gaps	Keep the platform and feed continuously connected; verify the data connection
Live curve inverted versus backtest	Look-ahead bias or bar-close vs intrabar mismatch in the strategy logic	Stop the strategy and re-test the logic in Market Replay

If the strategy is replicated across several accounts by the Trade Copier, review the master account first to measure the strategy itself, then review a copy account to measure replication quality and copy slippage.

Best Practices

• Review at fixed times. End of day, end of week, month-end. Scheduled review produces comparable reports; opportunistic review produces cherry-picked windows.
• Always set the Account filter. Never review with a blank account filter unless you deliberately want everything. Sim, demo, live and funded results must stay separate.
• Do the total pass first, then the per-instrument pass. The blank instrument filter gives the aggregate; single-symbol passes give attribution.
• Compare like periods. A full week against a full week, a calendar month against a calendar month. Comparing a good three days with a bad three weeks proves nothing.
• Keep the platform connected all session. Continuous history means continuous data. Starting and stopping the platform leaves data gaps that the AI Coach will call out and cannot interpret.
• Wait for 30+ trades before changing rules, and for 100+ before trusting profit factor and drawdown numbers.
• Turn AI Coach output into one measurable rule change, expressed as a number: max trades, max contracts, cut-off time, minimum stop distance. One change per review cycle keeps cause and effect visible.
• Read drawdown against your limits, not in isolation. Compare maximum drawdown and worst day with your daily loss limit and your funded account's trailing drawdown rule.
• Use Monte Carlo as a sizing tool. Size so the worst simulated drawdown still leaves the account compliant and tradeable.
• Pair the report with notes. The window tells you what happened; your journal tells you why. Together they produce a rule; separately they produce an opinion.
• Review losing periods first. They contain the actionable information. Winning-period reviews mostly confirm what you already believe.
• Keep the window on a separate monitor and closed during execution so live P&L does not drive in-session decisions.

Common Mistakes

• Leaving all filters blank and reading the result as "my performance". Blank means everything, including simulated and demo activity. In the demonstration the analysed trades were mostly demo trades taken while recording — statistically valid as a report, meaningless as a verdict on skill.
• Judging an edge on a single session. Twenty-seven trades in one day is a behavioural snapshot. Treat it as "did I follow my process?", not "is my system profitable?".
• Rewriting a whole strategy after one AI Coach suggestion. The coach names candidate problems; it does not authorise a rebuild. Make one change and measure it.
• Ignoring commissions and fees. Gross figures flatter high-frequency trading. A profit factor above 1 gross and below 1 net is a losing system.
• Confusing Lots with Trades. When you scale in and out, contract counts and trade counts diverge, and every per-trade metric computed from the wrong denominator is wrong.
• Treating Monte Carlo as a forecast. It is a distribution of possibilities derived from your own past trades, not a prediction of next month's equity.
• Reviewing only winning days. This systematically removes the information you need and reinforces recency bias.
• Comparing a live strategy run against a backtest without matching the date range. Including pre-deployment days or days the strategy was disabled makes the comparison invalid.
• Mixing instruments with different tick values in one report and reading the average trade in dollars. Compare in R-multiples or run separate per-instrument passes.
• Blaming a strategy for a sizing problem. If maximum drawdown breaches your limit but expectancy is positive, the edge is fine and the contract count is not.
• Never reconciling with the broker statement. If the report and the statement disagree, find out why (unrealised positions, fees, session boundaries) before trusting either.

Frequently Asked Questions

Where do I open the Trade Performance window in SabrTrader?

Open the SabrTrader Control Center, click New, and select Trade Performance. It opens as its own detachable window, so it may appear on a different monitor than the Control Center; drag it by the title bar to the screen you want and resize it to widen the filter fields and tab bar.

How far back can Trade Performance analyze my trade history?

The date range selector reaches back across multiple years. It offers Today, a trailing 12-month window, prior calendar years, and a fully custom start/end range. The practical limit is how much recorded trade history exists for the selected account: days on which the platform was not running or not connected produced no fills and therefore appear as data gaps rather than as flat days.

What happens if I leave the account or instrument filter blank?

A blank filter includes everything. A blank Account field reports all accounts together — sim, demo, live and funded — and a blank Instrument field reports all symbols together. This is useful for a total picture but dangerous for conclusions, because it blends environments with different risk realities. For any judgement about your live edge, set the account filter explicitly.

What is the difference between the Trades tab and the Lots tab?

The Trades tab lists completed round-turn trades: fills are grouped into one entry-to-exit trade, so a scaled entry and a scaled exit still count as a single trade. The Lots tab breaks the same activity down to the contract level, showing per-contract execution detail and total volume. Use Trades for win rate, expectancy and profit factor; use Lots to audit scale-ins, verify size and compute profit per contract. Ten trades of three contracts equals ten trades and thirty lots — never mix the two denominators.

What does the AI Coach actually analyze when I click Explain My Results?

It analyses only the trades currently selected by your date range, account and instrument filters. Effectively, the filters are the prompt. From that dataset it reports the trade count, what is statistically strong, what is weak, what stands out as anomalous, concrete items to work on such as risk management controls, and any limitations it cannot interpret — most commonly data gaps caused by the platform being started and stopped during the day.

Does the AI Coach only work on losing days?

No. It works on any filtered dataset, winning or losing. On a winning period it identifies which behaviours produced the result, for example a good win rate combined with a tightly controlled losing side. On a losing period it is more valuable, because it names the mechanism of the loss — loss size, trade frequency, timing or instrument — which is what a rule change needs to target.

Can I use Trade Performance to evaluate an AlgoStudio Pro or Strategy Builder strategy?

Yes, and it is one of the main use cases. Dedicate an account or a symbol to the strategy, set a custom date range matching the exact live run window, then compare the live equity curve, maximum drawdown, average loss and trade frequency against what the backtest predicted. Run the Monte Carlo tab to decide whether the observed drawdown is normal variance or genuine degradation, and use Explain My Results for a written summary of behaviour changes.

What does the Monte Carlo tab tell me about my trading?

It reshuffles and resamples your historical trade results many times to produce a range of alternative equity paths. That range shows how much of your actual outcome was sequence luck, what the worst plausible drawdown looks like at your current position size, and the approximate risk of ruin. It is a distribution of possibilities derived from your own trades, not a forecast of future profit.

Why does the report show data gaps?

Trade history is recorded while the platform is running and connected. If SabrTrader was closed, disconnected, or restarted during the session, no fills were captured for that interval, so the timeline contains gaps. The AI Coach flags such gaps explicitly because it cannot distinguish "no trades taken" from "trades not recorded", and period aggregation on the Periods tab becomes unreliable.

Why do my Trade Performance numbers differ from my broker statement?

Four common reasons: (1) Trade Performance counts realised, closed trades while an open position's unrealised mark-to-market is excluded; (2) commissions, exchange fees and other per-side costs may be accounted differently; (3) session and timezone boundaries can place a trade in a different calendar day than the broker does; (4) a filter is narrowing the sample — check the account, instrument and date range before assuming an error.

Can I separate simulated trades from live trades in the report?

Yes, using the Account filter. Select the specific account you want and the report excludes every other account. Because a blank account filter includes all accounts, sim and live results are merged by default — always set the account filter before drawing conclusions about live performance.

How many trades do I need before the statistics are meaningful?

Under 10 trades supports only behavioural review (did I follow my rules?). 10–30 trades reveals recurring patterns such as a weak hour or a weak symbol. 30–100 trades supports provisional expectancy and win/loss ratio and justifies one rule change. Above 100 trades, profit factor, drawdown profile and Monte Carlo output become reasonably stable. A single day of 27 trades is a snapshot, not an edge.

Can I move the Trade Performance window to a second monitor?

Yes. It opens as an independent, detachable window and can be dragged to any monitor. It may open on a different screen than the Control Center; simply drag it back. A dedicated review monitor or virtual desktop is recommended so the analytics never cover your charts and order-flow tools during the session.

How often should I run a performance review?

Run three cycles: a short end-of-day review (Today filter, live account, Equity Curve, Drawdown, Distribution, then Explain My Results), a longer end-of-week review (trailing week, per-instrument passes, Periods tab, Monte Carlo), and a month-end or yearly review using the trailing 12 months or a prior calendar year. Fixed review times keep reports comparable and prevent cherry-picking favourable windows.

What is the difference between the Equity Curve tab and the Drawdown tab?

The Equity Curve tab plots cumulative realised profit and loss, so it shows the direction and smoothness of your results. The Drawdown tab plots the decline from each running equity peak, so it shows the depth and duration of the giveback periods. The equity curve answers "am I making money?"; the drawdown chart answers "how much pain is required to make it, and does that fit my loss limits?".

Does the AI Coach require an AI provider or API key?

Yes. The AI Coach sends the filtered trade statistics to a configured large language model provider, so a provider and API key must be set up in SabrTrader's AI settings first. If the coach returns nothing, verify the provider configuration and key before assuming a data problem.

Can the AI Coach change my strategy or place orders?

No. The AI Coach is read-only analysis: it produces a written review of past trades. It does not modify strategies, alerts, orders or settings. Any suggestion it makes must be implemented deliberately — as a personal rule, an alert condition, or a change in a strategy built in AlgoStudio Pro.

Why does my equity curve look good but Monte Carlo look bad?

Because your actual trade sequence was one favourable ordering of many possible orderings. Monte Carlo reshuffles the same trades and exposes worse sequences: for example all the losses arriving consecutively. If the reshuffled worst-case drawdown breaches your account limits, the edge may be real but the position size is too large for it.

Troubleshooting

The Trade Performance window opened on another monitor or off-screen

The window is detachable and reopens on the display it last used. Drag it back by its title bar. If it is entirely off-screen, use the operating system window controls (right-click the taskbar entry and choose Move, or use the keyboard window-snap shortcuts) to pull it onto the active display, then resize it and leave it there so the position is remembered.

No trades are shown even though I traded

Check the filters in this order: (1) the date range — a Today preset excludes yesterday's session, and a custom range may end before your trades; (2) the Account filter — you may have selected an account with no fills; (3) the Instrument filter — you may have selected a symbol you did not trade. Clear the account and instrument filters temporarily (blank includes everything) to confirm data exists, then reapply filters one at a time. Also confirm the account actually had fills, not just working orders, in that window.

History is missing or gapped for part of the day

Fills are captured while the platform is running and connected. Closing, restarting or disconnecting SabrTrader during the session leaves gaps. Keep the platform and its data connection up for the whole session. When reviewing a gapped day, treat period aggregation and consistency statistics as unreliable and rely on the Trades tab for the trades that were captured. The AI Coach will explicitly note that it cannot coach around data gaps.

Numbers differ from my broker or prop firm statement

Reconcile in this order: confirm no positions were still open at the report boundary (unrealised P&L is excluded from the report); confirm whether commissions and exchange fees are included in the figures you are comparing; check timezone and session boundaries, since a trade closed after the broker's daily cut-off may fall in a different day; and verify that no account or instrument filter is narrowing the sample. Persistent, unexplained differences on a specific date usually indicate a data gap on that date.

The AI Coach returns thin, generic analysis

The coach can only describe the dataset you selected. A very small sample (a handful of trades) or a very narrow filter (one symbol on one day) leaves it nothing specific to say. Widen the date range to a week or month, or clear the instrument filter, and run Explain My Results again. Gapped history also reduces the depth of the review.

Explain My Results produces an error or no output at all

The AI Coach depends on a configured large language model provider. Verify that an AI provider and a valid API key are set up in SabrTrader's AI settings, that the provider account has available credit, and that the machine has internet access. If the key was recently regenerated or revoked, enter the new value and retry.

The window is slow to load a multi-year date range

Long ranges require processing every recorded fill. Narrow the date range (one year, one quarter or one month at a time), set the Account filter to a single account, and where possible set the Instrument filter. Build long-run conclusions from several smaller, faster reports rather than one very large one, and keep the Monte Carlo tab for the specific window you actually want to simulate.

Trade counts look wrong because I scale in and out

Trade Performance reconstructs round turns, so multiple entry and exit fills belonging to one position are reported as a single trade on the Trades tab. If you expected one row per fill, use the Lots tab, which shows the contract-level breakdown. Do not compute win rate or expectancy from lot counts.

Sim and live results appear mixed in one report

This is the blank-filter behaviour: an empty Account field includes all accounts. Select the specific live or funded account in the Account filter and re-read the report. If sim and live trades ran on the same account, they cannot be separated retrospectively — dedicate separate accounts going forward.

Glossary

Account segmentation

Filtering statistics per trading account so simulated, live and funded results are never blended into one report.

AI Coach

The panel in SabrTrader's Trade Performance window that, on pressing Explain My Results, sends the filtered trade statistics to a configured large language model and returns a written review of strengths, weaknesses, anomalies and concrete action items.

Algo / strategy performance review

Evaluating an automated strategy's actual live results over a defined window instead of relying only on its backtest output.

Backtest vs live divergence

The difference between simulated strategy results and actual filled performance, caused by slippage, latency, partial fills, commissions and thin liquidity.

Blank-filter behaviour

SabrTrader's convention that an empty filter field includes all values. A blank Account filter includes every account and a blank Instrument filter includes every symbol.

Commissions and fees

Per-side transaction costs charged by the broker and exchange. They reduce net performance and must be included before a profit factor or expectancy figure is trusted.

Daily loss limit

A hard stop on losses for a single session, either self-imposed or mandated by a prop firm or funded-account programme. Compare it with the worst day in the Periods tab and the deepest decline in the Drawdown tab.

Data gaps

Periods where the platform or its connection was not active, leaving incomplete trade history. Gapped history distorts period aggregation and is explicitly flagged by the AI Coach as uninterpretable.

Distribution skew

Which tail of the trade distribution is longer. Positive skew (long right tail of large wins, short left tail of controlled losses) is the target profile for most discretionary and trend systems.

Drawdown

The peak-to-trough decline in equity, used to size risk and judge whether a strategy is tolerable to trade.

Equity curve

A running cumulative plot of account profit and loss over time or trade sequence. Its shape reveals whether results come from a repeatable process or from isolated outlier trades.

Expectancy

The average expected profit or loss per trade, computed as (win rate × average win) − (loss rate × average loss). A high win rate with small wins and large losses can still produce negative expectancy.

Instrument-level attribution

Separating performance statistics per symbol so it becomes visible which markets actually produce profit and which are subsidised by the others.

Lots / contracts traded

Total volume executed, expressed in contracts or lots, as opposed to the number of completed trades. Used to compute per-contract efficiency and to reconcile commissions.

Maximum drawdown

The largest peak-to-trough equity decline recorded in the selected period. It is the primary input to position sizing and the primary constraint in funded-account rules.

Monte Carlo simulation

Randomised reshuffling and resampling of historical trade results to estimate the range of possible equity paths and worst-case drawdown. It separates edge from sequence luck and is a sizing tool, not a forecast.

Objective performance feedback

Data-driven review that removes recency bias and emotional interpretation of results by weighting every trade in the sample equally.

Overtrading

Taking more trades than the edge supports. Its signature in post-trade analytics is a high trade count with a flat or declining equity curve and rising commission drag.

Position sizing

How many contracts or shares are risked per trade relative to account equity and stop distance. Most apparent strategy failures found in post-trade review are sizing failures rather than edge failures.

Profit factor

Gross profit divided by gross loss. Values above 1 indicate a net-positive system; values between 1.0 and 1.2 are fragile once commissions are included.

R-multiple

A trade result expressed as a multiple of the initial risk taken on that trade. Risking $100 and making $250 is +2.5R. R-multiples normalise results across different position sizes and instruments.

Realized vs unrealized P&L

Realized profit and loss is the result of closed trades and is what performance statistics measure. Unrealized profit and loss is the mark-to-market value of an open position and is excluded from the report.

Risk management controls

Rules that cap damage on bad days, such as a daily loss limit, a maximum contract count, a maximum number of trades per session and a mandatory stop on every entry.

Risk of ruin

The statistical probability of losing a defined portion of account equity given current position sizing and measured edge. It rises steeply with size and falls with expectancy.

Round-turn trade reconstruction

Grouping individual fills into matched entries and exits so that partial scale-ins and scale-outs are measured as one complete trade rather than many.

Sample size

The number of trades inside the filtered analysis window. It determines how much any statistic can be trusted: under 30 trades supports behavioural observation only, while 100+ trades produces reasonably stable profit factor and drawdown figures.

Time-of-day performance

Breaking results down by session hour or session segment to identify when an edge is strongest or weakest, so trading can be restricted to productive windows.

Trade distribution

The frequency histogram of trade outcomes, showing how wins, losses and outliers cluster. A tight loss cluster with a long right tail of wins is the desirable shape.

Trade journaling

Recording trade context and outcomes and reviewing them so experience becomes a repeatable process. Trade Performance supplies the objective data; written notes supply the reasoning.

Trade performance analytics

Post-trade statistical review of closed trades to measure edge, consistency and risk. In SabrTrader it is delivered by the Trade Performance window.

Win rate

The percentage of closed trades that finished profitable. It is meaningless without the average size of wins versus losses.

Win/loss ratio

Average winning trade size divided by average losing trade size. A value above 1 means the average winner is larger than the average loser.

Related Documentation