There's a strange thing about back-testing an investment strategy: the easy part feels like the hard part, and the hard part is invisible.
The easy part is the mechanics — pick some instruments, set the timing, choose how the capital is allocated, press run, and watch an equity curve draw itself. Our AI-Strategy Builder does all of that for you: you decide what to invest in, when the strategy acts, and how the money is deployed, and it runs the entire historical simulation — every entry, every exit, every position — and hands you the result.
The hard part is a question almost nobody asks: can you actually trust the number it gives you?
Most of the time, the honest answer is "less than you think." And the reason is a bias so quiet that it flatters nearly every backtest you'll ever see online — including ones built on good intentions.
The hindsight trap
Here's the trap, and it's worth slowing down for.
Suppose you want to test a strategy over the last ten years. You sit down today and hand-pick a basket of strong companies to test it on. Reasonable, right?
Except you're picking from the list of companies that survived to today. The ones that went bankrupt, got delisted, merged away or quietly collapsed aren't on your screen anymore — so they never make it into your basket. You've unconsciously stacked the deck with winners you already know won.
This is called survivorship bias, and its effect is not small. Studies of exactly this problem find that leaving out delisted companies can overstate annual returns by one to several percentage points a year — and in some cases dramatically more — while quietly flattering your Sharpe ratio and hiding your true drawdowns. One analysis noted that a decade of North-American market data, if you only keep what's still trading, can be missing the majority of the companies that actually traded during that period.
Read that again: a backtest that only knows about today's survivors isn't a slightly optimistic backtest. It's a different, easier game than the one you'd actually have played.
The uncomfortable part is that this is the default almost everywhere. Most retail tools — and most "I backtested this myself" spreadsheets — simply don't carry the companies that died. You can't trade a delisted stock, so the data feeds quietly drop it, and your strategy is judged on a market that never existed.
The fix: rank the market as it really looked
The professional answer to this is something called a point-in-time (or dynamic) universe. Instead of choosing a fixed list today, the engine walks forward through history and, at each rebalance date, asks: "Given only what was knowable then, what would have ranked at the top?" — and it includes the companies that later faded or were delisted, exactly as they appeared at the time.
That's precisely what the Strategy Builder's automatic basket does. Flip one switch — "Build the basket from a Market Explorer ranking" — and instead of hand-picking instruments, you choose a lens and let the platform re-rank the market at every rebalance date:
- TyPulse™ — our all-round price-strength score: strong, steady, well-supported trends rather than the most explosive ones.
- Momentum, Sharpe, 1-year and 3-year return — purer performance lenses, for following whatever has been rising.
You set how many names to hold (the top N), how often to re-rank (every 1, 3, 6 or 12 months), and which day of the month the new list is drawn. Then the engine rebuilds the universe period by period, picking the top names as they ranked at that moment in history — losers and all. No hindsight. No quietly-deleted failures. A backtest of the game you'd actually have played.
This is the kind of thing that, until recently, lived only in expensive quant platforms. It's the feature that separates a backtest you can lean on from a backtest that's just telling you what you want to hear.
The loop that makes it click
Where this gets genuinely useful is how it connects to the rest of the platform.
The same rankings that drive the automatic basket — TyPulse, Momentum, Sharpe, 1Y and 3Y return — are the ones you can browse right now, live, on the Market Explorer. Same formulas, every asset type. So the workflow becomes a loop:
- Explore. On the Market Explorer, see which names rank strongest today, read the market's mood, watch where the money is rotating.
- Backtest. Take that exact ranking into the Strategy Builder and ask the only question that matters: "would following this lens, disciplined and rebalanced, actually have worked — through history, including the names that died?" The actual buy/sell/hold decisions on each instrument come from TyBuff's universal signal engine — the same asset-agnostic AI that drives every strategy, manual or automatic.
- Preview. Before anything runs, Preview basket schedule shows you the actual instruments selected in each historical period. Full transparency, no black box.
You're no longer guessing whether a screen "works." You're watching a rule play out across real, point-in-time history — and then you can save it as a portfolio that keeps re-ranking forward, freezing each period as it passes so your past results never silently rewrite themselves.
The honest part — because it matters more here, not less
We could stop at "look how powerful this is." We won't, because the people who care about survivorship bias are exactly the people who can smell a sales pitch.
So, plainly:
- Removing survivorship bias removes one lie, not all of them. A backtest is still a model of the past. It can't see regime changes, and a rule tuned until it looks perfect on history is its own kind of self-deception. Point-in-time data makes the test fair; it doesn't make the future known.
- A bigger back-test number is not a better strategy. A pure-momentum basket often posts the highest raw return in a rising market — and pays for it with a far rougher ride. Always read the drawdown and the risk-adjusted result, not just the final value.
- The engine-built history is capped at the last eight years. That's deliberate: coverage of delisted instruments thins out the further back you go, so we keep the window where the point-in-time data is most reliable rather than overreaching for a longer, shakier one.
None of that weakens the case. It is the case. A tool that shows you the losers, caps itself where the data gets thin, and reminds you that the past isn't the future is a tool built to make you a better investor — not to sell you a fantasy equity curve.
Where to find it
The Strategy Builder is yours to use today. The point-in-time dynamic basket — the automatic, survivorship-bias-free engine described here — is part of TyBuff Pro, alongside the full Market Explorer. It's the most serious piece of kit in the product, and it's built for people who want their backtests to tell the truth.
The fastest way to feel the difference: build the same strategy twice — once on a basket you hand-pick from today's winners, and once on an automatic, point-in-time ranking. Watch the two equity curves diverge. That gap, the one between the comfortable story and the honest one, is the whole reason this feature exists.