Backtesting software with fee and slippage estimates

What is backtesting, and what can it actually tell you?

Backtesting is the practice of running a fixed set of trading rules over historical price data to see how those rules would have behaved. It answers one narrow question well: was this idea internally coherent over the period tested? A strategy that loses money on its own historical data is not going to be rescued by live markets, so a backtest is a cheap way to discard bad ideas before they cost anything.

What a backtest cannot do is predict. It is a simulation of one past, under one set of cost assumptions, on one data set. Treating its output as an expected return is the single most common mistake in algorithmic trading, and the rest of this page is largely about the guards AKQX puts around that mistake.

How AKQX backtests: data, timeframes, fees and slippage

Data. The research database holds about 41.5 million (4.15 crore) historical candles, collected from public exchange and broker feeds covering crypto and Indian markets. That data is used for backtesting inside AKQX. It is not sold and it may not be copied or redistributed.

Timeframes. Every backtest can run from 1 minute up to 1 day, the same range the charts cover, so a rule you tested on a 15-minute candle is the rule you watch on a 15-minute candle.

Costs. Estimated trading fees and slippage are applied to each simulated trade rather than being left out. This matters more than it sounds: a strategy with a small average win per trade can look profitable before costs and be clearly unprofitable after them. Position sizing is fee-aware for the same reason, and a fee-to-risk breaker exists to stop setups where the cost of trading is large next to the risk being taken.

How do you read an AKQX backtest report?

The report is deliberately wider than net profit, because net profit on its own hides how a result was earned. Alongside net result, win rate and trade count, a backtest reports:

• Expectancy — the average result per trade. A strategy with expectancy at or below zero has no edge to scale, whatever its win rate looks like.
• Profit factor — gross wins divided by gross losses. Below 1.0 means the losses outweighed the wins.
• Maximum drawdown — the deepest fall from a peak, in percent. This is the number that decides whether a strategy is survivable, not whether it is profitable.
• Average win and average loss, plus win and loss streaks, which show whether a result came from steady behaviour or from a few outliers.
• MAE and MFE — how far trades moved against you before working, and how far they moved in your favour before you exited. These are what tell you whether a stop is too tight or a target is leaving money behind.

Reading these together is the difference between “this made money” and “this made money in a way I could actually have sat through”.

Why do backtests mislead, and what does AKQX check for?

Most backtests that look excellent are measuring the tester, not the market. The three classic failures are worth naming:

• Overfitting. Parameters tuned until the curve looks good on this exact history. The result describes the past precisely and the future not at all.
• Look-ahead bias. The rules quietly use information that would not have been available at the moment of the decision, such as a candle's close while that candle is still forming.
• Too few trades. A handful of trades cannot distinguish an edge from luck, however good the numbers look.

AKQX scores a result against these directly. A backtest is auto-rejected when expectancy is zero or negative, when profit factor falls below 1.0, when maximum drawdown exceeds 30 percent, or when the walk-forward, Monte Carlo or overfitting checks score below their minimum. Trade count has a floor too: 100 trades for crypto and forex, and 30 for Indian markets, where higher timeframes simply generate fewer signals. The Monte Carlo check reshuffles results across 1,000 simulations and treats a 30 percent drawdown from peak as ruin, which is a blunter and more useful question than “what was the average return”.

The algo scorecard and its verdict bands

Each backtest is scored out of 350 points, spread across seven areas: profitability (50), risk (50), consistency (40), quantitative checks (60), hedge-fund-style checks (50), intelligence (50) and execution quality (50). Weighting risk and robustness as heavily as profitability is intentional — it makes a fragile but profitable-looking result score poorly.

The total lands in one of six bands:

• Rejected — 0 to 112
• Review Required — 113 to 168
• Paper Trading Ready — 169 to 224
• Live Trading Ready — 225 to 280
• Professional Grade — 281 to 315
• Institutional Grade — 316 to 350

A band is a description of how one simulated test scored against AKQX's own thresholds. It is not advice, not a recommendation, and not a statement about what the strategy will do next. A result labelled Live Trading Ready has passed a set of checks; whether you ever deploy it with real money stays entirely your decision.

From backtest to paper trading, then your own decision

A backtest is the first of three steps, not the last. The second is paper trading: the same rules running on live prices with virtual capital and broker-style margin, which surfaces everything history cannot — real spreads, rejected orders, latency, and the gap between a theoretical fill and an actual one. Comparing paper results against the backtest that preceded them is the most useful single check available, because the drift between the two is the honest measure of how much the simulation was flattering you.

Only after that does real execution enter the picture, and only through your own exchange account with your own API keys. AKQX never holds funds and places no order you have not chosen to deploy. For Indian equities there is no real-money execution at all — charts and paper trading only.

Risk controls carry across each step: a maximum leverage cap, fee-aware sizing, a minimum stop distance, circuit breakers that halt trading after a set number of losses, and a daily loss cap. You can read more about turning discretionary concepts into testable rules or about the order flow behind each candle.

What a free account lets you test

A free account includes 10 backtests a month, 3 saved strategies, 1 paper algo and 5 AI requests a month, with one symbol and one timeframe per backtest. That is enough to test a real idea end to end before paying anything. Paid plans raise the monthly backtest allowance to 100 on Starter, 500 on Pro Trader, 1,000 on Advanced and 2,000 on Professional. Every price is shown in full on the pricing page before you pay.

Questions about backtesting

Related pages