What each strategy is waiting for on the chart, where it buys, and how it gets out: stop, target and trail, drawn to the real parameters. Each one ends with its actual journal record, good or bad.
Shares
US equities through Alpaca, plus US and LSE names on the Trading 212 ISA. QSR runs on every account. The two intraday models run only on paper-main.
QSR: buy the dip, sell the swing
qsr · daily · long-only
The signal comes from the sibling yahoo-screener: a quality name pulls back toward its 50-day average with RSI in the 30s–50s. This system doesn't buy the signal on sight. It waits for the dip to stop falling, buys a 2% bounce, and sells into the typical 4–6% swing.
live-1t212-isapaper-mainPaper02qsrPaper03QSR
1
Signal, then wait for +2% off the low
daily · 15-min poll
The screener flags the dip (signal). Instead of catching the knife, a watch opens. In trailing mode (live-1, t212, Paper03) the trigger follows the lowest low: trough × 1.02. When price trades through it, the normal entry runs. In fixed mode (Paper02) the trigger is the signal price × 1.02. The watch expires after 5–6 days if no bounce comes.
Each buy gets a real stop at 2×ATR, capped at 8% below entry, from the moment it fills. The anchor sells at a fixed +5%. On accounts with trailing turned on (t212-isa: 25% of the buy), the runner has no fixed target. Once it's up +6% a trailing stop arms and follows 3% below the peak. Elsewhere, the whole buy exits at +5%. No time stop: a QSR position is held until the target or the stop fills.
The record so far: strong on paper and the ISA, losing on live-1
The best-evidenced strategy here, with 500+ trades on paper-main. On real money the ISA is +£78.88 realized, FX-aware, across both books. live-1 is the problem child: the same signals lose there, with a profit factor under 1 after 137 trades. That gap is worth explaining before any capital moves.
VWAP reversion
vwap-mr-v1 · intraday · 2-min poll
Intraday, price that stretches too far below the session's volume-weighted average price tends to snap back. Buy the stretch; sell at VWAP; flat by the close.
paper-main
1
Stretched 0.5×ATR below VWAP, sell back at it
1-min bars · after 09:45 ET
When a bar closes more than 0.5× the daily ATR below VWAP (highlighted), enter. The target is VWAP at the moment of the signal, fixed rather than moving. The stop sits 1× ATR below entry. Anything still open at 15:55 ET is flattened, so nothing is held overnight.
session VWAPtarget = VWAP at signalentrystop 1×ATR
The record so far: positive, but thin
paper-main: 51 trades · 52.9% win · +0.17%/trade · PF 2.02. That beats its own 60-day backtest (−0.12%/trade, PF 0.87), which is a reason for caution as much as cheer. The sample is too small to trust yet, and profit factor on small trades swings hard.
Opening-range retest
orb-retest-v1 · intraday
Don't chase the opening breakout. Wait for it to come back and test the level. A false breakout can't hold a retest, so it never triggers.
paper-main
1
Break, retest, hold, enter
15-min range → 5-min · 120-min window
Box the first 15 minutes. A 5-minute bar closes above the range high. Price pulls back, dips to the level and closes back above it (highlighted): enter on the next bar. Stop goes under the retest low and the target is 2× the risk.
opening rangeentrystop (retest low)target 2R
The record so far: not living up to its backtest
Backtest: 432 trades, PF 1.50, positive in 3 of 4 windows. Paper so far: 36 trades · 19.4% win · −0.04%/trade. Most exits are near-flat end-of-day closes, and its reported profit factor (27) comes from one outlier, so treat it as noise. Keep watching; it hasn't earned more.
Retired share models
Not assigned to any account. Listed so they aren't re-invented.
swing-dip-v1
RSI < 40, 5–15% under MA50, rising volume. Bracket +5% / 2×ATR stop, 10-day time stop. Backtest +0.70%/trade; now superseded by QSR.
orb-v1
Buy the naked opening-range break on 2.5× volume. 68 paper trades, 45.6% win, ~0.00%/trade, PF 0.92. The breakouts kept reversing, so it was pulled.
quick-flip-v1
Fade a failed open. Rejected on a robustness test: the edge came from a few names and a single regime.
Crypto
24/7 through Alpaca on the shared crypto watchlist, no market-hours gate. Alpaca rejects bracket orders for crypto, so every entry is a plain buy followed by a separate stop_limit, and trails are moved by the script's own poll. Paper fills are fee-free; live costs roughly 0.4–0.5% a round trip. Judge live results after fees.
Trend: Donchian breakout
crypto-trend-v1 · daily bars · 2-hour poll
The Turtle Traders' rule: buy a new 20-day high, get out on a new 10-day low. Slow, few trades, and it lets winners run.
live-1paper-main
1
Break the 20-day high; trail from +15% on the daily close
daily · 2-hour poll
Enter when price clears the 20-day high (upper channel). The stop starts at 2×ATR, capped at 15%. Once a daily close is +15% up, a trail arms 5% below the peak daily close. It widens to 15% once the peak gain reaches +30% and 25% from +60%, so a runaway move isn't shaken out by its normal 10–30% pullbacks. The trail moves only on the daily close (00:00 UTC): moving it on every 2-hourly price chased intraday spikes and sold positions on ordinary noise. The stop itself still fires any time, and a new 10-day low sells whatever the trail is doing.
20-day high / 10-day lowtrail arms at a +15% daily closetrail 5% under the peak close (widens with gain)stop 2×ATR
The record so far: the best numbers here, on the fewest trades
paper-main: 10 closed · 70% win · +4.77%/trade · PF 4.82, with an average hold of 13 days. live-1 hasn't closed one yet: 9 open, about +1.4% unrealized. Ten trades is under the 20-trade floor, so this is promising, not proven. At this size of winner, fees barely matter.
Scalp: hourly breakout
crypto-scalp-v1 · hourly bars
The trend rule compressed to hours: a new 20-hour high in, a tight 2% trail out. Many more trades, each one smaller.
paper-mainlive-1 (off since 27 Sep)
1
20-hour high in; 2% trail after +5%
1-hour · max hold 2 days
Same mechanics as the trend strategy on hourly bars: break the 20-hour high, stop at 2×ATR capped at 10%. Where the trail is switched on (live-1 only; it's off on paper-main), it arms at +5% and follows only 2% under the peak, so it banks a spike fast. A new 10-hour low or 2 days held also ends it.
The record so far: great on paper, lost money on live
The clearest case of paper flattering a strategy: fee-free fills and a very different win rate. live-1 lost −$25.16 net on 34 trades and was switched off for new entries on 27 Sep. Its last open position (AVAX) is winding down.
Revert: Bollinger dip
crypto-revert-v1 · hourly bars
The opposite bet to the breakouts: when a coin falls out of its Bollinger band and RSI says oversold, buy the snap back toward the middle.
paper-main
1
Close below the lower band, RSI < 35
1-hour · max hold 2 days
Bands are a 20-hour average ± 2 standard deviations. A close below the lower band with RSI14 under 35 (highlighted) triggers a buy. The target is 60% of the way back to the middle band. The stop is 2×ATR, capped at 8%, and the position is closed after 2 days whatever happens.
Bollinger mid / bandstarget (60% to mid)entrystop 2×ATR
The record so far: small edge, and fees would likely erase it
paper-main: 18 closed · 50% win · +0.35%/trade · PF 1.60. On paper that's positive. Live crypto costs about 0.4–0.5% a round trip, which is more than the whole per-trade edge. Keep it on paper until the gross edge is clearly bigger than the fees.
Tape: buy the flush into bids
crypto-tape-v1 · 1-minute + order book
The only strategy here that reads the live order book. On a sudden 1-minute flush, it checks whether real buyers are resting just below. If they are, it buys the bounce off them.
paper-main (assigned, no trades yet)
1
Flush −1.5% into a bid wall
1-min · max hold 6 h
It needs all of these at once: a −1.5% drop within 15 minutes; the book tilted to buyers within 0.5% below price (imbalance ≥ 0.25, spread ≤ 0.3%); and sellers no longer dominating the tape (≤ 55% aggressive sells in 5 minutes). The stop sits just under the bid wall (0.4% buffer, max 3%) and the target is +1.5%.
resting bid walltarget +1.5%entrystop under wall
The record so far: none
Built and assigned to paper-main, with 0 closed trades. Its conditions are strict on purpose. With a +1.5% target and live fees near 0.5%, it would have to win well over half its trades to be worth taking live.