Method
How the lab computes, so that anyone can check. The choices below apply to every strategy, to HODL and to DCA, without exception.
The data
- BTC/USD, from Bitstamp, one candle per UTC calendar day: open, high, low, close. History since January 1, 2012, updated every night with the completed day. Last day: November 21, 2026.
- Each version of the data has a number, here 2026-11-21.58be4ea3, shown under every result. The history is never rewritten: a day already recorded no longer changes, and the same calculation always gives the same result.
- The data is stored by Ratiqen and served with the site: your browser does not query any exchange.
The calculation
- No future information. A decision made at a day's close is executed at the next day's open, at the opening price.
- Fees and slippage on every order: 0.10% fees and 0.05% slippage by default, on the amount traded, when buying as well as selling. Both can be set in the lab.
- A strategy targets a share of the portfolio in BTC, between 0 and 100%. The portfolio is only rebalanced if the gap reaches 2 points, to avoid tiny orders.
- Indicators (averages, RSI, highs and lows) are computed on the whole history since 2012. On the first day of a backtest, a strategy is therefore already in the position its rules had chosen.
- The exact rules of each strategy are shown in plain words on its page and in the lab.
The two benchmarks
- HODL: all the capital in BTC at the open of the first day, then nothing more.
- Monthly DCA: the capital split into equal installments, one installment invested on the first day of each month. What is not invested yet waits in cash, to compare with equal capital.
The measures
- Final value: the value of the portfolio at the last close.
- Annual return: the total return brought back to one year, compounding included.
- Max drawdown: the largest fall from a peak, for example -50% if the value went from $20,000 to $10,000.
- Sharpe: the average daily return divided by its variability, annualized. The higher it is, the steadier the return has been.
- Calmar: the annual return divided by the max drawdown.
- Time invested: the share of days when more than 1% of the portfolio is in BTC.
The detail of a result
In the lab, below the measures, four views explain where the result comes from. They break down the same value curve: none of them changes a figure.
- Buys and sells: each order, placed on the BTC price on the date it is executed, at the open of the day after the decision. The price shown is the opening price; fees and slippage are counted separately. The share of the portfolio in BTC is plotted day by day.
- Drawdowns: each day, the gap with the highest value reached so far. A drawdown runs from the peak until the value gets back to it; its duration counts the days spent below, up to the end of the period if the portfolio never got back.
- Year by year: the value of the portfolio at the last close of the year, compared with that of the previous year (with the starting capital for the first one). DCA keeps the part not yet invested in cash: its yearly return applies to the whole capital, as for the others.
- Sensitivity to settings: the same backtest rerun for seven values of each chosen parameter, from half to double the current setting, rounded to the parameter's step. The other parameters, the filters, the period and the costs do not change. Settings that make no sense, such as a short average longer than the long one, are left empty.
The verdict
Three questions, always against HODL and DCA: does the strategy end above, costs included? Is its max drawdown shallower? Does the result hold when starting from another date? For the last one, the backtest is rerun starting on January 1 of each year.
The safeguards
- The lab counts the variants tested. The more you try, the more likely the best one owes its place to chance. The count travels with the share link.
- It warns when there are too few orders for the number of conditions: the result then rests on a few decisions.
- It can hide the most recent 30% of the period while tuning, then reveal the verdict on that part: a result that does not hold on unseen data is not worth much.
- It shows the sensitivity to settings: a good isolated result, surrounded by bad ones, is probably luck. Settings computed by the map count as tested variants, and the map never looks at the hidden part.
The limits
- One candle per day: the order of prices within the day is unknown. All decisions are therefore made at the close, on the complete candle. When a strategy needs that order (a stop-loss, a price target), the least favorable scenario will be used. The grid, which does not see intraday round trips, is the least reliable strategy in the lab.
- Buy or wait only: no short selling, no leverage.
- Cash earns nothing. Taxes are not taken into account.
- Slippage is a fixed average: a large order, on a hectic day, would cost more.
- A single asset for now: bitcoin against the dollar.
The verification
The site's engine (version 0.2.0) is checked by more than a hundred automated tests. Its results are compared with those of a second engine, written separately in Python: on the seven strategies, from January 1, 2020 to October 12, 2026, both give the same figures.
What Ratiqen does not do
Ratiqen measures strategies on the past. It gives no signals, no recommendations and no personal advice: it measures, you decide.
Past performance is no guarantee of future results. Ratiqen is an analysis tool and does not provide investment advice.