What a backtest simulates

A backtest replays a strategy’s rules over past prices as if you had followed them. Every simulation makes assumptions, and the ones that flatter a result are rarely printed next to it. This page lists all of Tactfolio’s, so you can judge how much to trust what you see. For a shorter overview, see backtesting on Tactfolio.

Timing#

Each time a strategy runs, Tactfolio reads the rules using that day’s closing prices and trades at the same close. No rule can see a price from a later day.

Real trading can’t quite do this: you can’t know the exact closing price until the market has closed. So the equity chart adds a thinner orange Next-day open line: the same rules, trading at the next day’s opening price instead. Its CAGR and max drawdown lead the Reality checks table under the chart. If the two lines drift far apart, the strategy depends on trading right at the close. Every other number uses the same-day close. When a fund lacks opening prices, the line starts later or is left out.

Choosing the window#

The Window menu sets the dates: YTD, 1Y, 3Y, 5Y, 10Y, 20Y, Max (the default, all the history your strategy can use), or Custom dates. Your choice is remembered across strategies on this device. If a strategy’s history is shorter than the window you picked, the backtest uses what’s available without forgetting your choice.

Long windows are worth the wait. A strategy that looks great over five years may never have met a bear market; ten or twenty years usually include at least one.

Prices and data#

• Prices are daily and adjusted for splits and dividends, so returns include dividends as if they were reinvested.
• A backtest can only start once every ticker it uses has price history. The results show the dates actually covered, which can be shorter than the window you asked for.
• Missing data stops the backtest with an error instead of being filled in with a made-up return.
• The ticker list contains funds and stocks that trade today. Companies that failed or were delisted aren’t in it, which flatters any backtest that picks stocks (see survivorship).

Simulated history#

Some funds are young. The Simulated history option in the Window menu extends a young fund backwards with modelled values from before its first real price, so a strategy can be tested over a longer period. For example, GLD’s real prices begin in 2004 and its simulated history reaches back to 1968. Public strategy pages and classic strategy articles offer the same switch.

Simulated values are estimates, not prices anyone could have traded. The backtest switches to real prices from the fund’s first trading day and lists which tickers used simulated history. Treat results that depend on it with extra care. Simulated history explains how each fund is extended, the stand-ins used for asset classes older than any fund, and every ticker it covers.

Nominal and real returns#

A dollar in 1980 bought far more than a dollar today, so a strategy can look like it tripled your money while only keeping up with prices. The Nominal / Real switch under the headline numbers restates the backtest after inflation: each curve is divided by US consumer prices (CPI-U from the Bureau of Labor Statistics, and Robert Shiller’s monthly index before 1913), so $1 on the chart is $1 of purchasing power on the backtest’s first day. The metrics are recalculated; the strategy isn’t rerun, and turnover and allocations stay nominal. The note beside the switch names the latest CPI month used.

Real returns matter most over long windows and for comparing strategies across decades of different inflation. The workspace, Discover, and AI assistants’ results stay nominal.

Costs#

• Every trade pays slippage, 5 basis points (0.05%) of the amount traded. It is the same for every strategy, so results compare fairly; the Reality checks restate each run at 10, 20, and 40 basis points.
• Commissions, taxes, and fund fees beyond those already in the fund’s price aren’t modelled.
• Uninvested cash earns nothing. To earn a short-term Treasury return, hold a cash-like fund such as BIL.
• The Reality checks table repeats the same trades at 20 and 40 basis points (only levels above your own setting), so you can see how much a result depends on cheap trading.

Benchmarks#

The strategy is compared with a benchmark over exactly the same dates. The benchmark is a ticker, SPY by default, or another saved strategy. The starter patterns use QQQ, because it’s what their growth side could have held instead.

A strategy that takes less risk than its benchmark usually earns less and falls less. To compare at equal risk, the Reality checks table also shows the benchmark at the same risk: the benchmark scaled to the strategy’s volatility, with the rest in Treasury bills. It appears as a row named after the benchmark, such as SPY at same risk. Hover the row to see the size: at 0.72×, that’s 72% benchmark and 28% bills; above 1×, the extra is borrowed at the bill rate. If the strategy can’t beat it, holding the benchmark at that size would have done as well. These figures are nominal, so they show only in the Nominal view.

Metrics#

Reading a backtest explains how to use these numbers together.