00 / Performance
We would rather show you the method than a number we cannot defend.
Most platforms lead with a curve. We lead with how the curve is judged, because a curve is only as honest as the test behind it. This page shows exactly how we decide whether an edge is real, and reports the result in full: the crypto-only edge did not clear the bar, and we say so.
Paper and simulation only. No real capital. No claimed return.
01 / Method
Honesty is a process, not a promise.
Every result Canli Capital reports is produced the way the system would actually have to live. The model is trained on the past, tested on a future it has not seen, and rolled forward, with a purge and an embargo around every train and test boundary so no information leaks across it. There is one set of rules for research and one for reality, and they are the same set. Costs are charged in the test exactly as they are charged in paper, by a single transaction-cost authority. Signals computed at a bar's close are filled at the next open. Everything in the method is there to make the number trustworthy.
Out-of-sample, by construction.
Train on the past, test on a future the model never saw, then roll forward and repeat. A purge and an embargo sit around every boundary, so a label cannot peek across the seam into its own answer.
One cost authority.
Spread, fees, and funding are priced in the test exactly as the market would price them. The same authority that judges the backtest judges the paper fill, so research and reality cannot quietly diverge.
Decide at the close, fill at the open.
Every decision is made on a bar's close and filled at the following open. There is no looking ahead, by construction, so the test cannot trade on information it would not have had.
We test against ourselves harder than any market will.
02 / What we measure
Six ways to catch ourselves out.
When you search enough ideas, one of them will look brilliant by luck. The gauntlet exists to tell that apart from a real edge. Each test below asks the same hard question a different way: would this survive if the world had not been so kind? The tests use the following rigorous measures.
02 / The gauntlet
One question, asked six ways.
The same field of signals, run through every test in turn, until either one honest verdict remains or it does not survive the asking.
-
01 / Purged walk-forward
Out-of-sample by construction.
The backbone. Train on the past, test on an unseen future, with a purge and embargo around every boundary. If an edge exists only in-sample, this is exactly where it disappears.
Gate / positive and stable out-of-sample
-
02 / Deflated Sharpe Ratio
Discounted for how hard we looked.
A Sharpe ratio discounted for the number of trials behind it and for the non-normal shape of real returns. Given how hard we searched, how impressed should we actually be.
Gate / above zero after deflation
-
03 / Probabilistic Sharpe Ratio
A confidence, not a point estimate.
The probability that the true Sharpe sits above a chosen threshold, given the sample we have. A statement of confidence rather than a single flattering number.
Gate / high confidence above threshold
-
04 / Probability of Backtest Overfitting
The signature of a curve-fit.
Estimated by combinatorially symmetric cross-validation: the chance the configuration that looked best in-sample is no better than the median out-of-sample. Lower is better. A high reading is the classic tell of a curve-fit.
Gate / low, well below 0.5
-
05 / Combinatorially-purged cross-validation
Not one lucky path through history.
Performance estimated across many purged train and test combinations, so the headline figure is the behaviour of the method across histories, not the single kindest route through the past.
Gate / consistent across folds
-
06 / Must-beat-baseline gate
Beating zero is not the bar.
Every configuration we tried is counted and fed into the deflation, and the strategy must beat a simple honest baseline by a meaningful margin. A result that only beats doing nothing has not earned a dollar of risk.
Gate / beats a simple honest baseline by a meaningful margin
03 / The facts
What is true today.
These are the only numbers on this page, and every one is a property of the engine and its data rather than a performance claim. They are the facts you can hold us to.
- 0 Look-ahead, by construction
- 1 Cost authority, research and paper
- 2020 Crypto history starts
// Performance is deliberately absent from this ledger. AlphaForge has not traded real capital, so there is no live return to report, and we will not manufacture one from a backtest.
04 / Results
Here is the curve, and what it did not clear.
This is the validated out-of-sample result for AlphaForge: the equity curve, the Deflated and Probabilistic Sharpe ratios, the Probability of Backtest Overfitting, the trial count behind them, and the baseline it had to beat. It did not clear the bar, and we publish it in full. A results page that hides what disappoints is the one thing this company exists not to be.
This is a real, validated result. It did not clear purged walk-forward and the deflation, and we publish it in full. The path forward is breadth.
05 / Capacity
Size has a price.
An edge that exists at one notional may not exist at ten. The capacity curve traces expected return net of cost against deployed capital, so the point where added size begins to erode the edge through impact and funding is visible rather than assumed. Like every other number on this page, it is drawn from a validated artifact or it is not drawn at all.
The capacity curve is drawn from a validated artifact. It is dormant until one exists.
06 / Standing
The crypto-only edge has not cleared the bar. We are telling you on purpose.
Here is the honest status, the kind most platforms would never print. AlphaForge, tested rigorously on crypto perpetual futures alone, does not yet clear our validation gauntlet. The signal is not yet strong enough, after deflation and cost, to stake capital on. On this asset class alone it has not cleared the bar, so we report it rather than dress a backtest up as a promise. The differentiator of Canli Capital is not that we always win. It is that we run the test in the open and report what passes and what does not.
We would rather post a null than a number we cannot defend.
The deployed book pairs the crypto sleeve with a US equity momentum sleeve that does clear the bar (net Sharpe 0.91); the two are near-uncorrelated, and that is the edge. Breadth beyond the pair is the road forward. It is on the progress page.
07 / The current book
What the algorithms are holding, right now.
The house runs three algorithms: two standalone sleeves, AlphaForge (crypto funding carry) and AlphaMax (US-equity 12-1 momentum), and ALPHAC, their equal-risk combination, the flagship the firm reports on. In-sample validation reads 1.46, but after correcting for every strategy we tried it deflates to a central forward of about 0.7, with an honest range of 0.7 to 1.0. That is the number to lead with. Below is not a backtest but the live positions the book carries today, the real names, top per side, with the counts and the gross and net.
Honest forward Sharpe 0.7 to 1.0 in-sample 1.46
Holding cash, no open positions
The funding carry on crypto alone has not cleared the bar, so the sleeve is deciding to hold rather than stake capital on a signal that does not yet earn it. Holding cash is the position.
With AlphaForge holding cash, the cross-asset book is the equity sleeve today: . As the crypto sleeve earns its place, the book becomes the equal-risk pair, and the near-zero correlation between the two is the edge it is built on.
// These are the live paper positions, refreshed daily, drawn from the same ledger the engine acts on. No real money is in play. No name is shown that the book does not hold.