The four trading strategies — explained in full

Crypto Predictor runs four rule-based entry strategies on ETH-USD, BTC-USD and SOL-USD, each combined with one shared, fully automatic exit engine. This page describes exactly what the system does and when — the same rules run in paper trading and, later, in live trading. No black box, no AI guesswork at trade time: every entry and exit follows the deterministic rules below.

How the engine works

Market data. The system stores 15-minute and 1-hour candles from Coinbase, backed by a full year of history. Indicators are computed fresh on every evaluation — nothing is precomputed or curve-fitted at runtime.

Entry evaluations run every 15 minutes (at :00, :15, :30 and :45). For each product and each user, the user's chosen strategy is evaluated on the latest data. A new position opens only when all of the following are true:

1. The strategy's entry conditions (below) are met on this evaluation.
2. The user's position cap for that product is not reached (e.g. max 3 open positions per product).
3. Trading is enabled for the account and the product.

Entries fill at the live price. Signals are detected on the 1-hour candle close, but the moment a position opens it fills at the live market price at that second (plus fee and slippage) — exactly like a real exchange order. What you see recorded is the actual price you would have traded at.

Position sizing scales with your account. Your account has one total budget. Each position uses a fixed percentage of your current balance, recalculated every time — so a 25% size on a $1,000 balance opens $250 positions, and after the balance falls to $950 the next position is $237.50. Each position is sized at up to that percentage: how many you can hold at once is simply how many fit in your cash (25% ≈ 4), and the last one shrinks to use whatever cash is left (down to a $10 minimum) rather than being skipped. No martingale, no doubling down — exactly how a real account behaves.

Which coin gets funded first. When two or more products fire a signal in the same hour and there isn't enough free cash for all of them, the bot opens them in order. By default it backs the coin with the strongest current momentum first — the market moving hardest right now, measured by its recent 4-hour move. If you'd rather decide yourself, you can switch on a manual coin order in My Trading and rank your products explicitly (1 = funded first); momentum only breaks ties between coins you leave unranked. When cash is plentiful this never matters — every signal is funded — so it only comes into play on a nearly fully-invested account.

One entry per hour. While a strategy's conditions stay true, the bot opens at most one position per market per hour — it scales in gradually rather than stacking several correlated positions within the same hour. Combined with your position cap, this keeps each open position an independent bet rather than one oversized bet.

Costs are modeled honestly — 1:1 with real trading. Every simulated fill pays 0.05% fee per side plus 0.02% adverse slippage per side, and open positions pay (or receive) the real funding rate that OKX settles every 8 hours on perpetual futures — the exact rates the exchange publishes, applied to paper positions on the same schedule. Paper trading needs no exchange account — your account here is enough; all of this comes from public market data. Only if you later choose to trade real money do you connect your own exchange account (OKX perpetual futures; the app walks you through the setup in the backend). Because paper already charges everything real trading would, the results carry over 1:1.

The trend filter (used by three of the four strategies)

A market is considered in an uptrend when the price is above its 20-hour average, the 20-hour average is above the 50-hour average, and the 50-hour average is above the 200-hour average (price > EMA20 > EMA50 > EMA200 on 1-hour candles). The mirror image defines a downtrend. Anything else counts as no trend — and strategies that require a trend simply do not trade then.

The four entry strategies

The exit engine — identical for every strategy

Entries differ per strategy; exits never do. Every open position is checked once per minute against three protective rules, in this order:

1. Hard stop-loss: 3%. If the price moves 3% against the entry, the position closes immediately at market. This is the maximum planned loss per position.

2. Trailing stop: arms at +0.5%, trails by 0.25%. The system tracks the best price reached since entry (the peak). Once the position is 0.5% in profit, the trailing stop arms. From then on, if the price falls 0.25% below the peak (for shorts: rises 0.25% above the trough), the position closes — locking in most of the move instead of riding it back down. Until the peak keeps improving, the position simply stays open; the trailing level rises with every new peak and never moves backward.

3. Maximum holding time: 48 hours. A position that neither hit its stop nor armed its trailing stop within 48 hours is closed at market. Capital is never parked indefinitely.

Positions can additionally be closed manually at any time — per position or all at once per product.

A worked example (long): entry at $2,000. The stop-loss sits at $1,940. The trailing stop arms if the price reaches $2,020 (+0.5%). Say the price runs to a peak of $2,060 — the exit trigger now sits at $2,054.85 (0.25% below the peak). If the price pulls back to that level, the position closes with roughly +2.5% gross profit; if it keeps climbing, the trigger climbs with it.

What happens around every trade

• Every position gets a unique reference (paper: PT-XXXXXX, real: RT-XXXXXX) you can search in the app and quote in any question.
• If enabled in your profile, every open and close is pushed to your email or Discord within seconds — including entry/exit price, holding time, exit reason (trailing / stop-loss / max-hold / manual) and net P&L after all fees.
• A daily summary (optional, at your chosen time) recaps closed trades, open exposure, win rate and equity.

Honest limitations

• All backtest figures above are simulations over the 12 months to July 2026, per $300 budget and product, including fees and slippage. They are not promises. Market regimes change.
• The 12-month validation was performed on Bitcoin and Ethereum. Newer markets (e.g. Solana) run exactly the same rules but were not part of that validation window.
• Paper trading cannot fully reproduce real markets: order-book depth, liquidity gaps, funding costs and exchange outages only exist live. Expect real results to be somewhat worse than simulated ones.
• Automated strategies remove emotion, not risk. The stop-loss caps the planned loss per position — gaps or extreme moves can exceed it.

This page describes software behavior. It is not investment advice; see the Impressum for the full risk warning and liability terms.