The results page publishes every figure the programs produce. This page explains the markets being traded and how to read those figures. Nothing here changes the record — it is the same data, defined.

The markets

Each program trades one U.S. equity-index futures market, named by its exchange ticker:

YM — Dow futures

The E-mini Dow contract tracks the Dow Jones Industrial Average — 30 of the largest, most established companies in America, and the oldest continuously published U.S. market benchmark (since 1896). “The Dow” is the number the evening news reads; YM is how it trades.

NQ — Nasdaq futures

The E-mini Nasdaq-100 contract tracks the Nasdaq-100 — the 100 largest non-financial companies listed on the Nasdaq exchange, weighted heavily toward technology. It is typically the fastest-moving of the three markets.

RTY — Russell futures

The E-mini Russell 2000 contract tracks the Russell 2000 — two thousand U.S. small-cap companies. It is the standard benchmark for smaller American businesses and the broadest read on the domestic economy.

All three are standardized contracts listed by CME Group. They trade nearly 23 hours a day, five days a week, can be held long or short with equal ease (selling short requires no borrowing, unlike stocks), expire quarterly, and settle in cash — ongoing exposure simply rolls to the next quarterly contract.

How margin works on futures

Futures positions aren’t bought or sold like stocks — no money changes hands for the position itself. Instead, the exchange requires a performance bond, called margin, held on deposit for as long as the position is open. It is not a loan and accrues no interest; it is collateral that guarantees the position can absorb daily price moves.

Posting margin. Every open contract requires a set deposit determined by the exchange (brokers may require more). Long and short positions both post margin.

Day vs overnight margin. Brokers publish two rates. A position opened and closed inside one trading session posts the much smaller day (intraday) margin; a position held through a session close (5:00 PM ET) posts the full overnight (initial) margin. These programs source both rates from NinjaTrader’s published schedule and treat them as floors — the broker’s risk desk can raise intraday margins without notice when markets turn volatile. Every margin figure on the results page classifies each position day-or-overnight and applies the matching rate.

Marked to market. Gains and losses settle in cash at the end of every trading day. If cumulative losses draw the account below the maintenance level, the broker requires additional funds — a margin call.

Leverage. The deposit is a small fraction of the contract’s full notional value, so futures are inherently leveraged: gains and losses are amplified relative to the capital posted.

On the results page. “Margin in use” is the margin currently posted by a program’s open positions, shown as a percentage of that program’s reference account. “Peak posted margin” is the largest amount required at any one moment across the full record — a measure of how much capital the program actually ties up to produce its returns.

The reference account

Every percentage on the results page is measured against a fixed reference account — an account sized once for the margin each program requires, held constant through the record. Net results, annualized returns, drawdowns and margin figures are all fractions of it, computed without compounding. Dollar amounts are not published; the percentages carry the same information relative to account size.

Reading the performance figures

Legs. One completed position, from entry to exit. The published trade logs are the complete record — every closed leg, never truncated. Counts include each program’s validated record plus every live forward-test leg closed since, so they tick up on their own as the programs trade.

Win rate. The share of legs that closed profitable after costs. On its own it means little — a program can win often and still lose money — which is why it is paired with:

Profit factor, net of costs. Gross gains divided by gross losses. 1.00 is breakeven; 2.00 means the program made twice as much on its winners as it gave back on its losers. “Net of costs” means both sides are computed after modeled round-turn trading costs — commissions, exchange fees and slippage on every contract — so the ratio reflects what an account would actually keep, not a frictionless simulation. Sustained values above roughly 1.5 are generally considered strong.

Net · % of account. Cumulative net profit across the entire record, as a percentage of the program’s fixed reference account, computed without compounding. The “~%/yr” line beneath it is the same figure divided by the years in the record — the average yearly pace, with every flat month included in the denominator.

Net · on peak margin. The same net result measured against the peak posted margin — the most capital the program has ever had on deposit at one moment. Because the reference account is never fully deployed, this shows the return on the capital actually committed to the market; it will always read higher than the %-of-account figure.

Sharpe ratio. Return per unit of volatility — average monthly return divided by the variability of monthly returns, annualized (risk-free rate 0, as labeled). Months without a trade count as zero, so flat stretches lower the score rather than flatter it. Higher means smoother.

Sortino ratio. The same idea, but only downside variability counts against the score. It rewards programs whose bad months are rare and shallow, even if the good months are lumpy.

Max drawdown. The deepest peak-to-trough decline in realized equity, as a percentage of the reference account. This is the most the record ever gave back from a high-water mark before recovering — the number that tests whether an allocator could have lived through it.

Walk-forward efficiency. The anti-curve-fitting test. Parameters are chosen on one span of history, then performance is measured on data the selection never saw; WFE is the fraction of in-sample performance that survived out of sample. 1.0 means the results held up fully on unseen data; anything above roughly 0.4 is generally considered robust; above 1.0 means the rules did better on unseen data than on the data they were fitted to.

Sequence risk (MC). A pre-registered Monte Carlo exhibit: each program’s recorded trades are resampled into 100,000 alternate orderings (trades from the same price level kept together) and the equity curve is rebuilt each time. The figure shown is the 95th-percentile maximum drawdown across those orderings — how deep the same trades could have cut had they arrived in an unlucky order. “Ruin” means a drawdown so deep the program could no longer post its historical peak margin; zero of the 100,000 orderings reach it in any program. A sizing exhibit, not validation — validation remains the walk-forward results and the live forward tests.

Active now. Open positions marked to market, as a percentage of the account, alongside the margin those positions currently post. Unlike every figure above, this one is unrealized and moves with the market; it reads zero when the program is flat.

Measured in the open

The programs publish as they run. Closed legs are appended to the record and republished within minutes; the page shows its own data age honestly, including nights and weekends when markets are closed. Forward-test programs are executing their rules live against real market prices with simulated fills and costs — the record you see accumulates in real time, not in hindsight.