Permuter

History happened once. Your strategy was developed against that one sequence of bars, and every parameter you tuned was tuned to it — including the accidents. The 2024 range that happened to suit your stop distance. The gap that happened to land the right side of your entry. None of it will repeat.

This is why so many strategies look immaculate on historical data and die within weeks of going live. Not because the trader did anything obviously wrong, but because a single backtest is a sample size of one, and no amount of optimising a sample of one tells you whether the result was skill or coincidence.

Walk-forward analysis helps. Out-of-sample testing helps. But both are still carving up the same single history — and once you've looked at your out-of-sample data a few times, it isn't out-of-sample any more.

Permuter attacks the problem directly. It takes your real market data and manufactures alternative versions of it — histories that keep the asset's statistical personality but destroy the specific path your strategy memorised. Run your system across those. Survive them, and you have something real. Die on them, and you found out for the price of a tool instead of the price of an account.

And it works in the other direction too, which is the part most people miss. You can develop the strategy on synthetic data from the start — every idea, every parameter sweep, every failed experiment run against histories that never happened. Your real history stays sealed. When you finally test on it, that test is genuinely out-of-sample, because the strategy has never seen a single bar of it. You only get one first look at real data. This is how you keep it.

Used together they close the loop. Synthesise tells you the strategy works across markets that could plausibly have existed. Shuffle tells you it stops working when the structure is taken away — which is exactly what you want to see, because a system that profits on both is profiting on something that isn't there.

This is the part of Permuter you will not find anywhere else, and it's the reason the tool exists in the shape it does.

Most permutation tools operate on the whole file. You get a synthetic dataset or you get your real one. Permuter instead lets you define exactly which bars get touched — and every bar outside your selection stays bit-for-bit identical to source. Not resampled, not smoothed, not approximated. Identical.

What comes out is a single file containing both: stretches of genuine market history sitting alongside stretches that never happened, on one continuous timeline your platform reads as an ordinary symbol. Three ways to draw that line:

Slicing has an obvious limitation. If you permute the odd bands, the even bands are still your real, already-seen history — and half your timeline goes untested.

Chaining solves it. Select Both, and Permuter produces two complementary realisations from the same seed: one with the odd slices permuted, one with the even. Chained export then stitches the alternating chunks together into a single continuous series. Every region of your history appears in permuted form somewhere in that file — nothing is skipped, nothing is left as the version your strategy has already memorised.

The result is one long dataset that moves repeatedly in and out of unfamiliar territory, across the full span of your data, while preserving the asset's statistical character throughout. For training a model or stress-testing a system across regimes, that is a substantially harder and more honest test than any single permutation run — and considerably closer to what live markets do to a strategy.

Permuter is not a black box that hands you a file and asks you to trust it. Every configuration can be previewed on a full charting engine first — pan, zoom, crosshair, drawing tools, full-screen. Overlay as many permutations as you like against the real series and look at what you've actually built before a single byte is exported.

Every parameter is exposed and explained in the interface itself. Every run is seeded, so any result you get is reproducible exactly.

There is no refund policy, because there doesn't need to be one — you get the software before you pay for it, and you decide from there.

The free trial runs for 7 days and every feature is fully unlocked: both modes, all six methods, every permutation window, the full charting preview, and export. The single limitation is data volume — during the trial, each file you load is trimmed to a 30-day window that you choose yourself. Enough to evaluate the tool properly on your own data. Not enough to do a full validation run.

When you buy, you receive a licence file by email — permuter.lic. Install it and the limit is gone permanently.

Synthetic data is not real data. Permuter generates artificial and permuted price series for research and validation purposes. These series did not occur. Results obtained on synthetic data do not predict, guarantee, or imply any future market outcome.

No financial advice. Permuter is a software tool. It does not place trades, issue signals, recommend instruments, or provide financial, investment, or tax advice. Cave and Fire Group Pty Ltd is a software development company and is not a licensed financial services provider.

Validation is not a guarantee. A strategy that performs well across permuted and synthetic histories may still lose money in live trading. Statistical validation reduces the risk of deploying a curve-fitted system; it does not eliminate the risk of trading.

Your responsibility. Users are expected to understand the software's functionality before relying on its output, to interpret results with appropriate statistical caution, to maintain independent risk management practices, to ensure their use complies with all applicable laws in their jurisdiction, and to seek independent professional financial advice where appropriate.

By purchasing and using this software, users acknowledge that they have read and understood these notices and accept all risks described herein.