Skip to main content

Schism2MM Strategy: The Trough Hunter

Nickname: Trough Hunter, Rebound Harvester
Job: Trader who specializes in "buying the dip" after drops
Timeframe: 5 minutes (main battlefield) + 1 hour (big picture)

1. What Is This Strategy?​

Simply put, Schism2MM is:

  • A strategy that uses scipy scientific computing to find "price troughs"
  • Specifically enters at the bottom when others panic
  • Can even add positions! If you already have a position, you can buy more

Like an experienced trader calmly picking up bargains during a market crash 🏷️

2. Core Configuration: "Setting the Rules First"​

Take-Profit Rules (ROI Table)​

Just bought: Run at 2.5% profit
After 10 minutes: Run at 1.5% profit
After 20 minutes: Run at 1% profit
After 30 minutes: Run at 0.5% profit
After 2 hours: No profit required, watch for signals

Translation: The longer you hold, the more zen you become. Sell signals have your back anyway.

Stop-Loss Rule​

Maximum loss: 10%

Translation: Lose 10% and you're out, no negotiation.

3. Buy Conditions: I've Categorized Them for You​

This strategy's buy conditions come in two sets, I've grouped them into 2 categories:

🎯 Category 1: New Position (First Buy)​

Core Logic: Large timeframe up + small timeframe down + trough detected

In Plain English:

"Big brother (1 hour) is rising, little brother (5 minutes) is falling, and it's at the trough - time to pick up bargains!"

Representative Conditions:

  1. 1h RSI >= 65 → "Big trend is upward"
  2. Price <= 3-day low + 95% volatility → "Fell enough"
  3. RMI downtrend → "Indeed falling"
  4. Local trough detected → "This is the bottom"

Classic Lines:

  • Condition #1: 1h_rsi >= 65 → "Big trend is solid, safe to buy the dip"
  • Condition #2: close <= 3d_low + 0.95*adr → "Price is low enough, good margin of safety"
  • Condition #4: buy_signal == True → "scipy says this is a trough, trust it!"

📈 Category 2: Add Position (Already Holding, Want to Buy More)​

Core Logic: Held for over 3 hours + profit protection + trend upward

In Plain English:

"Already held for 3 hours, still rising, might as well add some more!"

Dynamic Threshold:

RMI threshold = linearly grows from 30 to 70
(Longer you hold, higher the requirement)

Classic Lines:

  • rmi-up-trend == 1 → "Trend is up, adding in the right direction"
  • current_profit > peak_profit * factor → "Previous profit still there, safe"
  • rmi-slow >= rmi_grow → "Dynamic threshold reached, can add"

4. Protection Mechanisms: 4 Layers of "Safety Fuses"​

Each buy condition comes with a set of protection parameters, like four security doors:

Protection TypePurposePlain English
Fixed Stop-lossExit at 10% loss"Lost too much, just run"
Dynamic Stop-lossHigher requirements over time"Held long enough without profit, leave"
Order TimeoutCancel if price deviates 1%"Price ran too fast, forget it"
Entry ConfirmationReject if price deviates 1%"Entry price wrong, better to miss it"

This is like checking three generations of family history before a date, extremely cautious 🕵️

5. Sell Logic: Even Flashier Than Buy​

5.1 Dynamic Stop-loss: How Much Loss Before Running​

Held 0 minutes: Allow 3% loss
Held 150 minutes: Allow 1.5% loss
Held 300 minutes: Require profit

Plain English:

  • Just bought: Small loss is fine, give some room for error
  • Held for a while: You should be profitable by now, if not then leave

5.2 RMI Stop-loss Signal​

ScenarioRMI ThresholdPlain English
Was profitableRMI < 50"Rose but now falling back, protect profit and leave"
Never profitableRMI < 10"Never got up, just take the loss and go"

5.3 Portfolio Profit Management​

This is the most clever part of this strategy - it considers overall portfolio:

if has free slots:
    Other positions avg profit >= -free_slots * 0.01  # Small loss is okay
else:  # No free slots
    Only allow biggest loser to sell  # Make room for new opportunities

Plain English:

"When there's room, be more lenient. When full, let the worst performer go first"

6. This Strategy's "Personality Traits"​

✅ Pros (Compliment Time)​

  1. Scientific Bottom Fishing: Uses scipy signal processing to find troughs, not guessing
  2. Can Add Positions: Good market conditions allow scaling up, not wasting opportunities
  3. Dynamic Thresholds: Stop-loss and add-position adjust over time, not rigid
  4. Portfolio Management: Considers overall positions, not fighting alone

⚠️ Cons (Complaint Time)​

  1. Too Complex: scipy signal processing? Linear growth functions? Headache to learn
  2. Backtest Difficult: Add-position logic, portfolio management can't be fully simulated in backtest
  3. Parameter Sensitive: order=100 extrema detection may fail in different markets
  4. Live Trading Only: Many logics depend on Trade object, backtest won't work

7. Suitable Scenarios: When to Use It?​

Market EnvironmentRecommended ActionReason
Oscillating UptrendDefault parametersStrategy designed for this
One-sided DowntrendDon't useTrend detection can't keep up, buying halfway down
High VolatilityBe carefulExtrema detection might get slapped
Sideways OscillationObserveMay trigger frequent stop-losses

8. Summary: How's This Strategy Really?​

One-sentence Review​

"An advanced player using scientific methods to buy the dip, but only for oscillating uptrend markets."

Who Should Use It?​

  • ✅ Technical types who know Python and signal processing
  • ✅ Traders who like buying dips and rebound trading
  • ✅ Players with live trading conditions for thorough testing
  • ✅ Experienced traders who can handle complex code

Who Shouldn't Use It?​

  • ❌ Newbies who get dizzy looking at code
  • ❌ Players only doing backtesting (many logics won't run)
  • ❌ Those who want to push hard in one-sided downtrends
  • ❌ Lazy people looking for simple strategies

My Suggestions​

  1. Dry-run first: Run in simulation for two weeks
  2. Understand scipy: Learn how argrelextrema finds troughs
  3. Careful with parameters: order=100 may not fit all coins
  4. Watch the trend: 1h RSI >= 65 is a hard requirement

9. What Market Can This Strategy Make Money?​

9.1 Core Logic: Trough Rebound​

Schism2MM is a buy-the-dip strategy. About 200 lines of code - what does that mean? Equivalent to a short academic paper 📚

Its profit philosophy: Precise entry during pullbacks in an uptrend

  • Scientific Extrema Detection: scipy finds troughs, not guessing
  • Trend Filter: 1h RSI ensures big direction is right
  • Dynamic Management: Longer holding means higher requirements

9.2 Performance in Different Markets (Plain English)​

Market TypePerformance RatingPlain English Explanation
📈 Oscillating Uptrend⭐⭐⭐⭐⭐This is the strategy's home turf, pullback buy then sell on rise
🔄 Sideways Oscillation⭐⭐⭐☆☆May repeatedly hit stop-losses, fees eat profits
📉 One-sided Downtrend⭐⭐☆☆☆Trend filter can't keep up, buying halfway down the mountain
⚡ High Volatility⭐⭐☆☆☆Extrema points jump around, stop-losses may get blown through

One-sentence Summary: Makes money in oscillating uptrends, loses money in one-sided downtrends.

10. Want to Run This Strategy? Check These Configurations First​

10.1 Trading Pair Configuration​

Configuration ItemRecommended ValueComment
Timeframe5m main + 1h infoDon't change to 1m, signals will be chaotic
order parameter50-150Use smaller values for high-volatility coins
Stop-loss-0.08 ~ -0.12Adjust based on tolerance

10.2 Hardware Requirements (Important!)​

This strategy's computation is okay, but has scipy signal processing:

Number of PairsMinimum MemoryRecommended MemoryExperience
1-10 pairs2GB4GBSmooth
10-30 pairs4GB8GBNormal
30+ pairs8GB16GBRecommended

Warning: scipy signal processing may eat memory with large data 😅

10.3 Backtest vs Live Trading​

Backtest Limitations:

  • Trade.get_trades() not available in backtest
  • Add-position logic cannot be fully simulated
  • Portfolio profit management fails

Recommended Process:

  1. First understand the code logic
  2. Dry-run for at least two weeks
  3. Small capital live test
  4. Observe add-position effects

Don't go all-in from the start, no matter how good the strategy, it needs breaking in!

11. Easter Egg: The Strategy Author's "Little Thoughts"​

Looking carefully at the code, you'll find some interesting things:

  1. scipy Signal Processing:

    "Using scientific methods to find troughs is much more reliable than eyeballing it"

  2. Linear Growth Function:

    rmi_grow = linear_growth(30, 70, 180, 720, open_minutes)

    "After holding 3 hours, RMI threshold slowly rises from 30 to 70, giving enough time for profit to fly"

  3. Portfolio Profit Management:

    "Lenient when there's room, let the worst performer go when full - this is veteran money management"

  4. 1% Price Protection:

    "If entry price deviates more than 1%, don't play - better to miss than to be wrong"

12. Final Words​

One-sentence Review​

"Advanced strategy using scipy for scientific bottom-fishing, a money-making tool in oscillating uptrend markets."

Who Should Use It?​

  • ✅ Quant players with technical skills
  • ✅ Those who like buying dips and rebound style
  • ✅ Have live testing conditions
  • ✅ Rational types pursuing scientific methods

Who Shouldn't Use It?​

  • ❌ Python newbies
  • ❌ Players only doing backtesting
  • ❌ Lazy people wanting simple strategies
  • ❌ Warriors pushing hard in one-sided downtrends

Manual Trading Suggestions​

If you want to follow manually without a bot:

  1. Add RMI(21,5) and RMI(8,4) indicators on TradingView
  2. Set alerts for 1h RSI >= 65
  3. Watch for local trough patterns
  4. Set 10% stop-loss on entry

13. ⚠️ Risk Re-emphasis (Must Read)​

Backtest Looks Great, Live Trading Needs Caution​

Schism2MM's backtest might look okay - but there's a trap:

Many core logics (add-position, portfolio management) can't run in backtest at all!

Simply put: Backtest can't test the real effect

Hidden Risks of Complex Strategies​

In live trading, complex logic may cause:

  • Add-position timing: May miss the best add-position point
  • Portfolio management: May misjudge across multiple coins
  • Extrema detection: May fail when market structure changes
  • Slippage impact: 5-minute timeframe is sensitive to slippage

My Advice (Sincere Words)​

1. Dry-run for at least two weeks, observe signal quality
2. Small capital live test for add-position effects
3. Watch if 1h RSI trend is accurate
4. Adjust order parameter for different coins

Remember: Scientific bottom-fishing is still bottom-fishing. How many have bought halfway down the mountain? Light position testing, survival is most important! 🙏