ALLOCATION ENGINE / HISTORICAL RESEARCH ARCHIVE
SEVEN YEARS.
THE FULL PICTURE.
SIMULATED · OPTIMISER-SELECTED
A historical view of AIE’s allocation engine, through rising markets, corrections and a severe drawdown.
March 2019–March 2026. This is a retained research result, separate from today’s native crypto bot platform.
Explore the monthly results ↓The strongest configuration was selected from 4,500 combinations over this period. These figures are historical simulation results, not actual customer returns or an independently validated forecast.
MONTH BY MONTH / NO MONTHS HIDDEN
THE GAINS.
AND THE LOSSES.
All 84 published monthly returns, preserved to two decimal places. Select a month to inspect its result, or filter by year.
PositiveNegative— Outside period
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2019 | — | — | — | |||||||||
| 2020 | ||||||||||||
| 2021 | ||||||||||||
| 2022 | ||||||||||||
| 2023 | ||||||||||||
| 2024 | ||||||||||||
| 2025 | ||||||||||||
| 2026 | — | — | — | — | — | — | — | — | — |
Select any month for its exact published return.
46 positive months / 38 negative months
On a small screen, swipe the results horizontally. Use the year buttons to focus on one row.
READ THE ASSUMPTIONS
WHAT WAS
ACTUALLY TESTED.
The figures below describe the original published report. The underlying run has not been rerun or independently audited as part of this website update.
2,560 trading days
The original report describes daily historical prices across a seven-year market period, including the 2020 shock and the 2021–2022 downturn.
Fees and slippage
Reported assumptions: 0.26% taker fee and 0.1% slippage per simulated trade. The report lists 142 trades and £65,169 in fees. Actual execution costs can differ.
4,500 configurations
The result is the selected optimiser winner on this period. Selecting the strongest historical configuration creates selection bias; this archive provides no separate out-of-sample result.
Approximated reasoning
Rules and historical market inputs approximate the allocation engines. This does not replay historical live AI reasoning, and is not a native Revolut bot backtest.
UNDERSTAND THE NUMBERS
RETURN IS
ONE PART.
A strong ending value does not remove the risk taken to reach it. Read the loss periods and assumptions alongside the headline.
67.5% maximum drawdown RISK
The largest reported fall from a previous peak. It represents a severe decline within the simulation, even though the full-period result was positive. It is not a maximum possible future loss.
1.327 Sharpe · 1.344 Sortino RATIOS
The report gives these measures of return relative to overall volatility and downside variation respectively. They are retained as reported; calculation inputs and annualisation should be checked against the original research before drawing comparisons.
52.5% positive days FREQUENCY
The old report calls this “win rate”, but defines it as days with positive returns. It is not the percentage of profitable trades. The separate monthly series contains 46 positive and 38 negative months.
Reported results and present-day use SCOPE
This research belongs to the historical allocation engine. It does not establish the performance of today’s GRID or DCA bots, a customer deployment, or the isolated public demo.
HISTORICAL EVIDENCE / CLEAR LIMITSSTUDY THE RESULT.
STUDY THE RESULT.
UNDERSTAND THE RISK.
Optimised historical results can overstate future performance. Market conditions, liquidity, costs and system availability change. Capital is at risk; a backtest is not a promise.
EXPLORE WHAT EXISTS TODAY