Three trades can each have a sensible stop and still put the same account in an unsensible position. The missing number is often not the loss on a trade. It is the loss already committed across all trades that are alive together.
Portfolio heat in prop backtesting is the aggregate loss the open portfolio would take if every active position reached its defined exit, including the costs the test says apply. It is a snapshot of committed downside, not a forecast and not a substitute for equity drawdown.
A closed-trade report can hide this completely. It records each result after the positions have stopped overlapping. A prop account experiences the overlap first.
Portfolio heat is not exposure, correlation, or drawdown
Portfolio heat is easy to confuse with several useful measures. They answer different questions, and a sound backtest keeps them separate.
| Measure | What it describes | What it can miss |
|---|---|---|
| Portfolio heat | The loss committed by the current open positions if their exits are reached | A gap, slippage, or an exit that cannot fill at its assumed price |
| Exposure | How much time or capital remains in the market | Whether the stops line up in the same adverse move |
| Correlation | How returns or price moves tend to co-move over a chosen sample | A short, shared shock and the actual stops open at that moment |
| Drawdown | The fall the account has already travelled from a reference point | How much open downside was loaded before the fall arrived |
The exposure guide is about vulnerability while trades remain open. The correlated-trades guide explains why several apparently separate entries can fail together. Portfolio heat joins those ideas at the account snapshot: it totals the risk that has already been accepted.
That distinction matters for a cBot with more than one symbol or strategy. A portfolio may be diversified over a long sample and still have several positions vulnerable to one currency move, one index sell-off, one spread shock, or one piece of scheduled news. Diversification is a property to test. It is not permission to ignore the open book.
Define the heat calculation before viewing the result
Portfolio heat is only useful when its contract is written before the backtest is inspected. At each account snapshot, record every open position and calculate its defined loss from the current exit policy.
portfolio heat = sum of each open position's loss at its defined exit
+ declared costs needed to exitFor a long or short trade with a hard stop, that normally means the difference between the current exit-relevant price and the stop, converted through the symbol's contract specification and position size. For a trade without a hard stop, the formula has no natural ceiling. Calling that risk "zero until closed" is not a calculation; it is an omission.
Write down the choices that change the number:
- whether the snapshot uses the stop price, a stress price, or a liquidation rule;
- whether pending orders count, and when they become committed risk;
- the spread, commission, swap and slippage assumptions at exit;
- how symbol volume, contract value and account currency are converted;
- whether the measure is shown in account currency, as a share of equity, or both;
- the platform timestamp and the prop firm's applicable rule clock.
The exact loss limits and reset conventions belong to the prop firm and can change. Keep the current rule source with the test; FTMO's Trading Objectives are a source for its own accounts, not a shortcut for every firm. The cTrader Symbol reference is likewise the place to check the live contract fields a cBot uses rather than guessing pip value or volume behaviour.
A portfolio heat peak is an event, not an average
An average heat figure can be reassuring for exactly the wrong reason. A prop rule is tested at a moment. The maximum concurrent heat, the time it occurred, and the positions that made it up are usually more useful than a tidy average across quiet bars.
Build a heat series alongside the equity series. For every bar or event at the resolution the test can genuinely support, preserve:
- account balance and equity;
- open positions, their symbols, direction, volume and defined exits;
- heat by position and total portfolio heat;
- open P/L, realised P/L, costs and any pending orders;
- the strategy or signal that created each position.
Then identify the peaks. A peak made by two independent positions on unrelated dates has a different meaning from a peak created by several entries after the same trigger. The report should let a reviewer see the difference instead of treating the total as an abstract risk score.
This is also why bar resolution matters. If positions can be opened, stopped and reopened inside one bar, an end-of-bar snapshot can miss the high-water mark of committed risk. The bar-close versus intrabar backtesting guide covers the wider execution problem. Here, the practical rule is simpler: do not claim to have measured a heat peak at a finer path than the data and engine recorded.
Test the account path, not a position in isolation
A position-level stop can be correct while the portfolio-level result is not. The test needs a ledger that applies entries, exits, fills and costs in chronological order and then asks what the account was carrying at every point.
Use a small, pre-written scenario set in addition to the historical run:
| Scenario | What to change | What the report should show |
|---|---|---|
| Concurrent stops | Several active positions reach their defined exits in the same interval | Total heat, realised path, and whether the account constraint is crossed |
| Spread expansion | Exit costs widen under a declared stress assumption | The difference between baseline heat and stressed heat |
| Pending activation | A pending order becomes live while existing heat is elevated | Whether the new risk is counted before or after activation, as specified |
| Gap through exit | The next executable price is worse than the stop assumption | The loss beyond stop-based heat and the affected account path |
| Correlated shock | Positions linked to the same driver move adversely together | Which positions created the heat peak and whether the dependency was visible beforehand |
The point is not to select a scary scenario after seeing the curve. Freeze the scenarios, the cost model and the reporting fields first. A strategy that survives a declared stress is not thereby safe. A strategy that fails one has given useful information about its constraints.
How portfolio heat changes a prop backtest
Prop-firm constraints turn aggregate open risk into an account problem. A final profit can be positive while the route to it includes a heat peak that would have put the account under pressure. Whether that peak breaches a rule depends on the current rule definition, the time zone, floating equity, costs and the actual execution path. It cannot be inferred from a monthly return table.
The funding model keeps the account path beside its constraints instead of reducing an evaluation to its endpoint. Use that frame when reviewing heat: compare the peak against the stated account conditions, and preserve the inputs so someone else can challenge the calculation.
At realbacktesting, the published methodology makes the data, execution, cost and validation assumptions visible. realbacktesting is a trading-software studio for cTrader built around inspectable, reproducible testing. That does not make a backtest a live track record. It does make the question sharper: what did the portfolio have at risk, under what assumptions, and could another trader reproduce the answer?
Frequently asked
Is portfolio heat the same as a stop-loss total?
Only if every open position has a defined stop and the calculation includes the declared exit costs. Portfolio heat is the account-level total at a moment; a list of stop distances without sizing, contract conversion and concurrent positions is not the same measure.
Can low correlation make portfolio heat safe?
No. Low historical correlation can help diversification, but it does not cap the loss already committed by the current positions. Measure both the relationship among trades and the open risk at the snapshot.
Should pending orders be included in portfolio heat?
The answer depends on the written risk contract. A pending order is not an open position, but it can become risk while the existing book is already under pressure. Report it separately at minimum, and state exactly when the model counts it.
Can a backtest prove a heat limit will protect a prop account?
No. It can show how a declared heat rule behaved on the tested data and scenarios. Rule changes, gaps, execution differences and unobserved intrabar paths remain outside that proof.
The stubborn takeaway
The risk that fails a prop account is often not hidden in a bad trade. It is sitting in the open portfolio, waiting for several ordinary trades to become one loss.