RSI+MACD Trend Strategy by grishkamm
By grishkamm
Performance Metrics
- Author: grishkamm
- Symbol: NASDAQ:AMZN
- Timeframe: 1 day
- Win Rate: 70.6%
- Profit Factor: 2.73
Description
Strategy descriptionConceptA long-only swing strategy for US equities that is built around one idea: the exit should be driven by trend invalidation, not by distance from the price peak. Trailing stops cut large moves short and shake you out on ordinary pullbacks. This strategy replaces them entirely — it holds through drawdowns inside a healthy trend and lets go only when the weekly trend structure actually breaks.The edge, if there is one, lives in the exit architecture, not in the choice of indicator. RSI and MACD are ordinary; the way they are combined across two timeframes is the point.How it worksThe strategy runs on two timeframes at once:Weekly — permission and invalidation. MACD(12, 26, 9) on weekly closes decides whether trading is allowed at all, and it is the only thing that can close the remaining position.Daily — timing and execution. RSI(14) picks the moment to enter inside an already-established weekly uptrend.A trade unfolds in three stages:Entry. While the weekly MACD line is above zero (trend is ON), a daily RSI(14) cross up through 45 triggers a buy at the next day's open. RSI 45 is deliberately not an oversold level — it marks a pullback inside a trend that has resumed, not a bottom-fishing attempt.Scale-out. The first 2 × ATR(14) of profit takes 50 % off the table via a resting limit order. This banks part of the move and halves the exposure of what remains.Runner exit. The rest is held — through pullbacks, through consolidations, with no trailing stop — until the weekly MACD line crosses below its signal line. That closes everything at the next day's open.If the position is stopped out or shaken out and the weekly trend is still ON, the same entry rule fires again. Re-entry is a feature, not an accident — it is the designed answer to false breaks, and it is why no trailing stop is needed.Risk modelThere is no conventional stop-loss in the core design. Position risk is governed by three things instead:Fixed cash per position (default $2 000). This is the real loss limiter — each ticker is an independent sleeve, so a $10 000 account can run five names.The scale-out, which removes half the exposure once the trade is working.An optional disaster stop (default ON, 4 × ATR below entry, fixed — not trailing). It is intentionally far away: it should stay silent in the vast majority of trades and only catch the tail, where the weekly MACD is too slow.Execution model — no repaintingEvery signal is computed on closed bars only:The weekly MACD is read as the previous completed weekly bar, so a forming week never influences a decision — in history or in real time.All entry/exit logic is gated behind barstate.isconfirmed, so nothing flickers inside an unfinished daily bar.A signal on a bar's close is executed at the next bar's open, matching how the backtest is computed and how you would actually trade it.This means the alerts are tradable: what you see on history is what you get live.What to expect (honestly)Trend-following gives back a meaningful part of the open profit at the end of every trend. That is the cost of capturing the full move, not a defect.The weekly MACD exit is slow. Sharp bear legs produce the worst trades, and that is exactly what the disaster stop is there for.The win rate is misleading. The scale-out is a separate closed trade that by construction only fires in profit, so it mechanically inflates win %. Judge the strategy by profit factor, max drawdown and worst trade instead.One parameter set is meant to work across all tickers. Per-ticker tuning is overfitting and is explicitly out of scope.