Bitcoin perpetuals · 5-minute chart

Most strategies show you their best year. Here are all five.

Selepro Veltus is a trend system for BTCUSDT.P. It made money in three of the last five years and lost money in two. Both are on this page, because the losing years are the part you actually need in order to decide.

See the record Live signal log Subscriptions opening soon

Net return by year

Regime-sized exposure, 25–150% of equity · 50% margin · 0.05% commission · 2 ticks slippage · bar close
−13.5%
2022
−4.5%
2023
+16.5%
2024
+36.7%
2025
+19.0%
2026*

*2026 to 12 August. Two consecutive losing years came first — one of them mild, one not. The worst drawdown across the whole period was 28.1%, and the equity curve spent about 35 months below its April 2022 peak before making a new one. The longest run of consecutive losing positions was seven. Compounded over five years the result is positive, but if you cannot sit through a stretch like that, this system is not for you — and it is better to learn that here than eight months into a subscription.

Before you subscribe

What this will feel like to hold.

All of the following is documented behaviour measured across five years, not worst-case speculation. Knowing it in advance is the difference between sitting through a drawdown and quitting at the bottom of one.

  • Losing years happen. Two of the last five: −13.5% in 2022 and −4.5% in 2023. The system makes money from sustained direction and loses money in extended ranges — that is the shape of the edge, not a defect awaiting a fix.
  • The worst drawdown was 28.1%, measured across the whole five-year period. Within any single calendar year the worst was 19.9%, also in 2022 — the deeper figure comes from 2022 and 2023 compounding on each other.
  • Recovery took a long time. The equity curve sat below its April 2022 peak for about 35 months.
  • Losses cluster. Across 306 positions in five years, runs of one to three consecutive losers are ordinary. Four and five happened five times each, six once, and seven once — the worst on record. Each position exits in two tranches, so a run of five losing positions appears in a TradingView report as ten consecutive losing trades.
  • The ten best trades of the year carry between 47% and 72% of that year’s gross profit, depending on the year. Most trades do very little, and long flat stretches are normal.
  • It wins slightly more often than it loses. 53% of positions across five years, ranging from 44% in 2023 to 66% so far in 2026.
  • Judge it over quarters, not weeks. A month tells you almost nothing about a system like this.
Validation

The tests that were designed to make it fail.

A backtest is easy to make flattering. These are the checks that separate a discovery from a curve fit — and what each one returned.

Lookahead audit

No lookahead found

Parameter robustness

7 of 8 perturbations stayed profitable

Walk-forward on sizing

Improved in both untouched periods

Cross-asset test on ETH

Failed — published, then found to be our error

Extended history

Found two losing years

A dozen modifications

All discarded
How the engine works, and what each test measured →
Scope

One instrument, one timeframe, honestly bounded.

Built and validated on BINANCE:BTCUSDT.P, 5-minute chart. Between 44 and 78 positions a year, typically held one to two days. Each position exits in two tranches — TradingView counts each tranche as a trade, so a report shows roughly twice as many.

It was tested unchanged on Ethereum across the same five years and lost or broke even in four of them. We published that, and then found out the test was ours to blame: it ran Bitcoin’s stop widths on an instrument that moves about a third more per bar. The full correction is here. What still holds is narrower and worth stating plainly: every market needs its own risk calibration, and not every market has an edge to find — the same procedure applied to Solana produced −31.1% and was discarded. Nothing published on this site has been measured on another timeframe or another exchange.

Live tracking

Every fill, timestamped on arrival.

Entries and exits are recorded the moment they arrive, each one cryptographically chained to the last. Editing history would break the chain, and a button on the page lets you recompute the whole thing in your own browser.

Recording since 12 August 2026
The log shows what the strategy did, not what it earned. A signal is not a recommendation to trade.
Open the live log →
Also from Selepro

Selepro Flow — free, and useful on its own.

A market regime indicator that answers one question: is price travelling, or is it churning? It came out of this research, where the strongest finding was that performance sorted almost entirely by how directional the market was — not by any entry filter. No signals, no predictions. Works on any symbol and timeframe.

Access

Selepro Veltus opens to subscribers soon.

The strategy is published on TradingView as an invite-only script. Subscriptions are not open yet — the live record started on 12 August 2026, and we would rather you judged it on real forward results than on backtests alone.

Subscriptions opening soon
Everything else is available now: the full record, the research behind it, the documentation and the live log. Nothing is behind a paywall except the strategy itself.
Read the documentation →
Want to know when it opens?
One email when subscriptions open. No newsletter, no drip sequence, and your address goes nowhere else.

Questions before then? Write to support@selepro.com, or message Selepro on TradingView.

Questions

The ones worth asking.

Why publish the losing years at all?

Because you would find out anyway, and finding out later is worse. A strategy that only ever shows good periods is either very young or very selective about its window. Knowing the bad stretches in advance is what lets you size sensibly and hold through a drawdown instead of quitting into one. These figures were corrected on 13 August 2026 after the published record was checked against the exported trade list and found not to match; the corrected version is what you see here.

Do I need an account on a specific exchange?

No. You need TradingView with data for BTCUSDT.P. All validation used Binance's order book, so that is the chart to run it on. Where you execute is your decision, and the fills and costs you get there are yours.

Can I use it on Ethereum or other coins?

Not with these settings. The published script is calibrated for BTCUSDT.P, and running it unchanged elsewhere is the mistake described in this note — the stop distances are the part that does not travel. An Ethereum calibration has been built and measured, and two validation checks on it are still open; no figures for it will be published here until both are closed, and the script you can subscribe to today is the Bitcoin one.

Do I have to watch the chart all day?

No, but signals arrive on 5-minute bar closes, so you need alerts running and a way to act on them within a reasonable window. Positions are typically held one to two days, not minutes.

When does access open?

Soon. The script is already published on TradingView and the live log has been recording since 12 August 2026. Subscriptions open once there is enough forward record to judge it on.