Strategy version: TradeAMW 1.0 · Test period: July 2021 to June 2026 (five years) · Data through: June 30, 2026 · Report build v351 (ID 26b01/e50e3) · Methodology last updated: August 1, 2026
What is a backtest?
A backtest applies a strategy's rules to historical market prices to calculate what those rules would have produced over the period studied. It is a historical test of how the strategy's rules would have behaved, not a record of actual trading. Backtests are prepared with hindsight and do not capture every real-world cost or condition.
The data
All performance figures were produced from CME E-mini S&P 500 (ES) futures market data over two windows: a five-year window, July 2021 through June 2026, and an extended ten-year window, July 2016 through June 2026. The site shows the five-year record. The five-year window includes the 2021 rally, the 2022 bear market, and the recoveries and rangebound stretches that followed, so the strategy was measured across rising, falling, and sideways conditions.
How the backtest was produced
The strategy's exact rules were replayed against that history, trade by trade. Every entry, stop, target, and exit was calculated using the historical prices contained in the test dataset, with no discretionary overrides. A commission of $4 per order is included in the results.
The current-tier model
Every performance figure on the main page comes from one calculation, the backtest at current tier settings.
Each trade in the research record carries its own stop distance, derived directly from the research model's position sizing. The model replays every trade in chronological order at the single 0.50% risk setting and applies, in order:
- Position sizing at 0.50% of the account. Before each trade, the intended risk is 0.50% of current account equity. The position is sized so that if the trade hits its stop, the loss equals that intended risk.
- A fixed ceiling of $20,000 of risk per trade. No single trade risks more than $20,000, regardless of account size or risk level. This ceiling is applied to every figure on the main page.
- Order costs. A commission of $4 per order is included inside every trade result.
- Per-trade compounding. Each trade's profit or loss changes the account balance before the next trade is sized.
- AMW fees. At the end of every calendar month a management fee of 2% of account value is deducted. A 30% performance fee is then deducted from any amount by which the account finishes above its high-water mark, the highest post-fee month-end value recorded so far. No performance fee is charged in a month that ends below that mark. The next month compounds from the post-fee balance.
This single model produces every figure on the main page: the equity curve, the monthly grid, the drawdown chart, the annual bars, the benchmark comparison, and every headline statistic.
Effective risk declines as the account compounds
The $20,000 per-trade ceiling has an important consequence that anyone reading these figures should understand. Because the ceiling is a fixed dollar amount while the account compounds, the risk actually taken per trade falls as a percentage of equity over time. Once an account is large enough that its stated percentage would exceed $20,000, the ceiling binds and the stated percentage is no longer what is being risked.
At the single 0.50% setting the ceiling begins to bind once account equity passes $4.0 million, because 0.50% of $4.0m is exactly $20,000. Across the five-year backtest 62% of all trades were capped by the ceiling rather than sized by the percentage. Measured year by year, the average risk actually taken per trade was:
| Year | Risk actually taken per trade | Largest decline that year |
| 2021 | 0.50% of equity | -23.0% |
| 2022 | 0.50% of equity | -25.0% |
| 2023 | 0.47% of equity | -22.6% |
| 2024 | 0.34% of equity | -20.6% |
| 2025 | 0.16% of equity | -10.5% |
| 2026 | 0.11% of equity | -6.5% |
Read the headline drawdown figures in this light. The deepest declines happen in the opening years, when the account is still small and the full 0.50% is genuinely being risked. That is not a level of risk maintained across the whole backtest: by 2025 the ceiling had throttled the average trade to 0.16% of equity, and the yearly declines fall with it. The early years are the honest guide, not the late ones.
An account that does not compound past $4 million, which includes any account starting near the stated minimum, keeps risking the full 0.50% throughout. Such an account should expect declines closer to the 2021–2023 figures above, roughly 23% to 25%, than to the mild later ones, and closer still to the 25% recorded across the five-year window.
Exact treatment
| Historical market | CME E-mini S&P 500 futures (ES) |
| Test period | July 2021 to June 2026 (five years) |
| Entry/exit prices | The strategy's backtested fill prices on historical data |
| Commission | $4 per order, included in trade results (an order = each entry or exit execution; a completed trade is at least two orders) |
| Slippage | Not modeled |
| CFD spread | Not modeled |
| Overnight financing | Not modeled |
| Position sizing | 0.50% of the current account per trade, capped at $20,000 of risk on any single trade |
| AMW fee | 2% of account value per month plus 30% of profit above the high-water mark, deducted at month-end |
| Compounding | Per trade, chronological |
| Deposits/withdrawals | None assumed |
| Open positions at month-end | Trades count in the month they close |
| Taxes | Excluded |
The two market regimes in the ten-year record
The ten-year record is not uniform, and the difference is the single most important thing to understand about it. Between 2016 and 2020, price moves followed through far less often than they do today. We measure this directly from the strategy's own closed trades as the ratio of average favorable movement to average adverse movement. That ratio sat between 1.35 and 1.57 in every year from 2016 to 2020, and between 1.70 and 1.99 in every year from 2021 to 2026. The two ranges do not overlap at any point.
This matters because a trend-following strategy earns its returns from follow-through. In the 2016–2020 regime the strategy remained profitable in four of those five years, but it earned a small fraction of what it earns today. That is the honest shape of the record, and it is why the five-year and ten-year views report very different annualised figures for the same unchanged strategy.
The current version of the engine measures this ratio continuously from its own last 400 completed trades. When follow-through falls below the historical threshold, the engine automatically tightens the thresholds that determine which setups qualify as tradeable, and it becomes more selective. This adaptation is applied on the basis of live measurement, not on the basis of dates.
Two limitations we state plainly
Follow-through is currently near a ten-year high. The strategy earns more when moves follow through and less when they do not. The 2021–2026 figures were produced in favorable conditions for this style of strategy. The 2016–2020 portion of the record shows what materially less favorable conditions look like: still profitable in most years, but far less so. A reasonable expectation for the future sits between the two, not at the recent end.
2020 is the only true volatility shock in the record, and it sits outside every window used to develop the strategy. In that year the strategy finished profitable, but with a within-year decline several times larger relative to its annual return than anything in the recent five years. Our expectation for a comparable future shock is a roughly flat year accompanied by a drawdown materially deeper than any shown in the five-year charts.
What counts as a trade
The five-year backtest contains 4,497 completed round trips (entry to exit) across the strategy's signals; the two-year backtest contains 1,880 and the ten-year backtest contains 8,275. 4,497 completed round trips means 8,994 individual executions: each round trip is exactly one entry execution and one exit execution, and trades with partial exits are recorded as separate round trips in this count. The $4 cost applies per execution and is already inside each trade's result.
Relationship to the research output
The unconstrained research output before fees is materially larger than the published figures. The figures offered publicly are the capped, fee-inclusive results shown on the main page.
The benchmark comparison
The benchmark chart on the main page compares the backtest at current tier settings at the selected risk level, after stated AMW fees, with the S&P 500 over the same period. The S&P 500 line is a total return series built from published annual total returns (dividends reinvested), Slickcharts. The strategy uses leverage and has a different risk profile from holding an index fund; the comparison does not make the two equivalent.
The live instrument
The current strategy trades a broker-issued index product that follows the price of the S&P 500 index. The exact product name and symbol vary by broker; on the platform it may appear under a label such as US500 or SPX500. These are leveraged contracts, not ownership of stocks or an index fund: the account gains or loses based on movements in the index price without holding the underlying shares. Broker pricing, spreads, financing charges, margin requirements, and execution quality affect results and vary by broker.
What ongoing monitoring covers
Alongside the strategy's automated controls, a person supervises each trading session. Monitoring covers:
- Connection status between the strategy, the alert bridge, and the broker.
- Platform errors or outages on the charting, bridge, or broker side.
- Rejected or unfilled orders.
- Unexpected or mismatched positions in the account.
- Daily loss control status (whether the circuit breaker has paused new entries for the day).
- The ability to pause the system if conditions require it.
If a technical problem occurs, the response process is to identify the issue, pause new activity when needed, reconcile the account's open positions against what the strategy intended, and resume only once the system is confirmed to be operating normally. Monitoring reduces operational risk; it does not remove market risk or guarantee that every issue is caught or corrected before it affects the account.
What is a contract for difference?
A contract for difference (CFD) is an agreement with a broker to exchange the difference between a product's price when a position is opened and its price when the position is closed. It allows the account to gain or lose based on movements in an index price without owning the underlying stocks. CFDs are leveraged: a smaller amount of account money controls a larger position, which magnifies both gains and losses. A CFD is a contract with the broker rather than an exchange-traded product, so the broker's pricing, terms, and creditworthiness apply. CFD prices and trading conditions may differ from the underlying cash index and related futures markets.
Limitations
Backtested results are hypothetical. They are prepared with hindsight, involve no financial risk, and do not model slippage or liquidity: fills are assumed at historical prices, which live markets do not guarantee. No client earned the displayed results, and no real money was traded in the test. Live execution takes place on a broker-issued index CFD following the S&P 500; the product name and symbol vary by broker, and CFD spreads, financing costs, and trading conditions differ from the futures data used in testing. Stop orders and daily loss controls are intended to reduce risk, but they do not guarantee a maximum loss. Live accounts also meet margin requirements and liquidity constraints that the model does not capture, Live results may differ materially from backtested results and may be less favorable.
Plain-language glossary
- Broker
- The regulated firm where the trading account is opened, funded, and held. Deposits and withdrawals happen at the broker.
- Strategy
- The fixed set of rules that decides when to enter, manage, and exit trades.
- Automated
- The software places and manages trades without manual order entry. It does not mean the account needs no oversight.
- Backtest
- A calculation of what the strategy's rules would have produced on historical market prices. Not live trading.
- Leverage
- Controlling a position larger than the money committed to it. Magnifies both gains and losses.
- Drawdown
- The decline from a previous account high. An account that grows to $100,000 and then falls to $80,000 is in a 20% drawdown.
- Stop-loss
- An order intended to close a trade after it moves against the position by a set amount. Not guaranteed to fill at its exact level.
- Performance fee
- A fee calculated as a percentage of trading profits, as defined in the client agreement.