How Blitz Works

What T-DCA actually does

T-DCA stands for Tiered Dollar-Cost Averaging. It is an execution framework, not a forecast. The bot does not decide whether the market will go up. It decides, in advance, how much a move against the position is allowed to cost.

A cycle runs in one direction:

1. First entry. The bot opens a position that is deliberately small relative to the capital assigned to the strategy. Being immediately wrong has to be survivable.
2. Tiered adds. If price moves against the position, the bot adds at planned intervals. The tiers are not uniform — the spacing and the size of each step come from the strategy template, and every step is known before the cycle starts.
3. Take profit. When the position reaches its take-profit target it closes in full, and the cycle resets — the ladder returns to step one with nothing carried over.
4. Repeat. The next cycle starts from the current price, not from the last one.

Two consequences of step 3 matter more than they sound. Drawdown is not cumulative across cycles, because a closed cycle leaves nothing behind. And the deepest tiers are reached rarely, because most cycles close long before the ladder is spent. How deep the ladder actually went, year by year, is published on the backtest page.

Every parameter above — entry count, tier spacing, take-profit target, leverage, stop-loss — is visible and editable in the strategy template once you are signed in. Choose from multiple ready-to-run T-DCA strategies, copy and edit one, or build your own. Backtest your settings free in the app, with no plan-gated lookback and no monthly run quota.

Why there is no trend filter

Blitz has no RSI, no moving average, no trend condition on entry. That is a deliberate choice, not a missing feature.

Indicators are conditional probabilities, not forecasts. They are right often enough to be useful and wrong often enough to be dangerous. The problem is not that a signal fails — it is what a trader does when they believe one. Confidence justifies a larger first entry, and the size of the first entry is what determines how far the ladder can still absorb a move against it. A filter does not remove risk; it moves risk out of the position structure and into the forecast.

T-DCA assumes the opposite. We do not claim to know direction. The first entry is sized small enough that being immediately wrong is survivable, and everything after that is handled by a ladder with a fixed number of steps and a fixed capital cap — not by a prediction.

There is a cost, and it is worth stating plainly. A filter would help in one specific market: a long, one-directional decline where every rebound is smaller than the take-profit threshold. In that market the ladder is exhausted and the position is no longer averaging down — it is waiting. We do not solve that with a forecast. We bound it in advance: a finite number of entries, a maximum share of capital that can ever enter, a stop-loss you define, and a first entry small enough to leave the ladder intact.

We do not try to predict when to enter. We decide in advance how much being wrong is allowed to cost.

Why positions run in Cross margin

Blitz runs futures positions in Cross margin, and there is no option in Blitz to switch this. On most supported venues the engine syncs the margin mode to Cross before a cycle starts. A small number of Hyperliquid assets are isolated-only at the venue itself; there the bot follows the venue rule.

The reason is the ladder. Under Isolated margin each position is walled off with a fixed slice of margin, and a move against an early tier can liquidate that slice while the rest of the account sits idle — the later tiers never get to do their job. Cross margin lets the whole assigned balance stand behind the ladder, which is what the tier schedule assumes when it plans how deep it can go.

This is a trade-off, not a free choice. Cross margin means a position draws on the account balance rather than a capped slice. That is why the first entry is small, why the number of tiers is finite, and why a maximum share of capital is fixed before the cycle starts. The measured worst-case distance to liquidation is on the backtest page.

What happens when the ladder runs out

The ladder is finite by design, so there is a point where it has no steps left. We would rather describe that than leave it implied.

When the last tier has been used, the position stops averaging down. From there it is a single position waiting for price to return to its take-profit level. The bot does not add beyond the planned tiers, and it does not raise the capital ceiling to keep going. If you have set a stop-loss, that is what ends the cycle instead.

This is the situation the design accepts in exchange for not relying on a forecast. How often it happened over the backtested period — and how close those cycles came to liquidation — is published with the rest of the risk figures on the backtest page.

What we do not control

Blitz executes on your own exchange account, which means parts of the outcome sit outside the bot.

• Your settings. Leverage, take-profit, stop-loss and the entry ladder are yours to change. A template that behaved one way in a backtest behaves differently once those values change.
• Your exchange. Order rejections, API downtime, funding rates, maintenance windows and liquidation rules belong to the venue. Rules also differ between venues — a stop-loss type available on one may not exist on another.
• Manual intervention. If you close or modify a position directly on the exchange, the bot sees the result, not the intent.
• The market. A backtest measures the past under stated assumptions. It is not a forecast, and it does not bound what a future market can do.

Funds never leave your exchange account and withdrawal permission is never requested — that is covered under Security. The limits above are about outcomes, not custody.