Using BLEEEP

BLEEEP is not a leaderboard of screenshots. It is a workbench where users build strategies and one fixed, public engine judges each one against the same bar for everyone. Every strategy is user-built; the engine never runs its own book. This page describes what users actually do with it.

What users do

• Build in chat. Describe a strategy to BLEEEPY in plain language: “fade funding when it goes past this level on BTC and ETH, size flat, no leverage, exit after N hours.” It compiles the description into a BleeepSpec, the frozen definition the engine runs. Users can read the compiled spec, edit any field, or write the spec directly. The natural-language step is a convenience, not a black box. The spec is the source of truth, and every line is inspectable before it runs.
• Test it before any capital is at risk. Run the frozen spec through the validation gate and get a GO, NO_GO, or INSUFFICIENT verdict with the metrics behind it, before any capital is at risk. A NO-GO indicates which specific test the strategy failed, making clear why it would have died, not just that it did. The user revises and resubmits. What cannot be done is move the passing bar: it is fixed and public.
• Set the terms. Choose the markets, the parameters, the risk limits, and the position sizing.
• Go from paper to live, safely. Paper forward-testing runs on real prices with no real orders, on the exact decision path live execution uses, so the paper record is comparable to live behavior, while still being clearly labeled as simulated. Promotion to live capital is a human decision, and live is off by default. When it is on, the Safety Vault sits between the strategy and user funds: session keys, spend caps, a destination whitelist, a circuit breaker, a daily-loss auto-pause, and reconciliation against the exchange.
• Subscribe to forward-verified strategies. Allocate to any user’s strategy on the strength of a sealed, on-chain track record rather than a +300% screenshot. A subscription is non-custodial: the subscriber receives the strategy’s live signals and deploys them on their own funds, through their own Safety Vault and a human deploy gate, never an automatic mirror. Browse listings in the strategy marketplace.
• Keep logic private, prove results. Every strategy’s decisions seal and anchor before their outcomes, so a user can keep their own logic private while the resulting track record stays provable to anyone.

The dashboard

From one place users watch live and paper strategies side by side, read the sealed track record and on-chain proof for any strategy they follow, see each strategy’s current verdict and the metrics behind it, and browse the NO-GO archive of everything tested and rejected. When a strategy trades, the sealed record can be pulled up and the math checked directly.

Markets

BLEEEP initially supports Polymarket prediction markets, including crypto-related binary markets, and is extending to on-chain perpetual DEXes such as Lighter, native on-chain and fully verifiable. The engine reads from a flexible data catalog rather than a hardcoded venue list, so support extends to new markets as their data is captured, without changing how strategies are written.

Why BLEEEP instead of an ordinary trading bot

A normal bot hands the user a backtest and asks the reader to trust it, then runs logic that cannot be inspected and reports numbers that cannot be checked. BLEEEP inverts both halves. A user controls and sees everything about their own strategy, and the one thing that cannot be touched is the standard a strategy is judged against, because a movable bar is not a test: the bar is fixed and public for everyone. The result is sealed and anchored, so the track record is provable to anyone.

Composer, QuantConnect, and similar tools let users build and backtest, and some run live. What none of them do is prove the record. Their performance is computed and displayed on their own servers, with nothing anchored, so a curve on their dashboard is exactly the “trust my number” the user was trying to escape. The difference is not more features. It is that the track record is verifiable by anyone, not asserted by the house.

Why this builds a stronger portfolio

Proof is the method. The goal is a portfolio users can trust enough to actually fund.

The single biggest way retail traders lose is not bad luck and not fees, though both hurt. It is deploying a strategy that looked brilliant in the backtest and quietly dies in live trading. The backtest was overfit, tuned knowingly or not to the exact history it was tested on, so it scored a beautiful curve on the past and had no edge on the future. The money goes in on the strength of that curve and comes out in a drawdown the backtest never showed.

BLEEEP’s transparency gate is aimed straight at that failure, and it runs before capital is committed. BLEEEP does not promise returns. It filters out strategies that fail the tests most fragile backtests cannot survive: deflated Sharpe, walk-forward out-of-sample, multi-regime durability, cost realism, and adversarial refutation.

What users build from the survivors is a diversified book allocated to proven track records instead of fake screenshots. Different markets, different regimes, different edges that each passed on their own merits, each checkable on-chain rather than rendered on someone’s server.

BLEEEP does not promise returns. It filters out strategies that fail the tests most fragile backtests cannot survive.

Preset Strategies

Not everyone wants to write a spec from scratch. Presets are ready-made BleeepSpec templates for common approaches: trend-following, mean reversion, momentum, breakout, volatility-based, and conservative risk-managed setups. Sensible defaults are already filled in. Picking one by preference, experience, market view, or risk appetite yields a working strategy in one click. Advanced users open the same preset and tune every field, so beginners get an easy start and experienced traders keep full control.

A preset is a starting point, not a promise. It still has to pass the same validation gate as any other strategy. This matters. Some classic approaches, such as short-horizon momentum and breakout on crypto, tend to be regime luck, and when a preset fails the gate, that is the product working correctly. It shows, provably, why a popular idea dies in live trading before it is funded. Unlike bots that brand a preset with a name and a promise, every preset’s track record here is validated and anchored like everything else.

Verifiable discovery (planned)

A discovery layer is a natural extension once the core is trusted. BLEEEP could surface potential opportunities from many signals: trading volume, unusual venue activity, funding and basis dislocations, sentiment shifts, news flow, and on-chain flows. This is a larger build and belongs after the verification core, as a separate module, not in the first release.

The framing is what keeps it on-brand. A discovery signal here is not a call the reader is told to trust. It is a candidate to verify. When BLEEEP spots an anomaly it proposes a hypothesis, seals it as a committed prediction before the outcome, and runs it through the same deterministic engine and transparency gates. Every alpha call is born as a sealed prediction, so its hit rate accrues on-chain and cannot be revised after the fact. Most will be NO-GO, and that is the verifiable denominator, the same archive that already backs the rest of the product.

This is the opposite of the usual “AI says buy, high confidence” signal service, which is unverifiable. In BLEEEP’s framing, a discovery signal is not a trade recommendation. It is a hypothesis that must be sealed, tested, and scored before users treat it as useful. Here the AI translates a hunch into a testable, pre-committed, scored hypothesis. The engine still judges. The AI only proposes.

Next: browse the strategy marketplace, see how the passing bar is defined in the validation standard, or how strategy privacy works in User strategies.