INCUBATOR v2 — AI TRADING

🔨 BUILDER

What It Is

Evolution of v1's winning strategy. Same mechanic (long spot, short perp, collect funding), but with 2-3x leverage on the short perp to amplify returns from ~1.5%/month to 3-7%/month.

Economics at Both Capital Levels

Why Leverage Changes the Game

In v1, the strategy was market-neutral but only returned 15-20% annually. The v1 Boardroom's Challenger correctly noted "I can do that with stocks." Adding 2x leverage on the perp short doubles the funding income without increasing directional exposure (the spot hedge still covers the full position). The risk: liquidation. If BTC dumps 50% in a day, your short perp profits but your spot loses. The hedge holds. But if BTC pumps 50%, the short perp needs more margin or gets liquidated. At 2x leverage, you survive a 40%+ pump before liquidation. At 3x, you survive ~25%.

Key Risk

Liquidation on the leveraged short. BTC pumped 30% in a single day (Nov 2024). At 3x leverage, that would have margin-called the position. At 2x, it survives but gets uncomfortable. The bot must auto-deleverage if unrealized loss on the perp exceeds 60% of margin.

Automation

Fully automated via Binance API. Bot monitors funding rates, enters when rate > 0.05%/8h, exits when rate drops below 0.01%/8h. Sends daily Telegram reports. Jarvis can run this 24/7.

Builder Verdict: USE WITH CAUTION ⚠️

🔴 CHALLENGER

The Liquidation Math the Builder Buried

The Builder says "at 2x leverage, you survive a 40%+ pump." Let's check. You short BTC perp at $60K with 2x leverage ($500 margin for $1,000 position at $1K capital). BTC pumps to $84K (+40%). Your perp loss: $400. Your margin: $500. You have $100 left. That's not "surviving" — that's one more 5% move from liquidation. And the spot hedge? Your spot BTC is worth $700 (up $200). Net: you're still down $200 overall during the pump, and your margin is nearly wiped.

The REAL liquidation threshold at 2x leverage is closer to 35-40%, not "40%+." And BTC moved 30% in a single day in November 2024. So the historical record shows this leverage level nearly gets blown up within the last 12 months.

Funding Rates Are NOT Consistent

The Builder assumes 0.05% per 8h average. Historical data from Binance shows:

• Bull market average: 0.05-0.15%/8h (great for this strategy)
• Bear market average: -0.01 to 0.01%/8h (strategy earns nothing or bleeds)
• Crash events: -0.1 to -0.3%/8h (YOU pay the funding, at 2x leverage)

In bear markets, this strategy doesn't just stop working — it reverses. You're paying funding instead of collecting it. With leverage, those losses are amplified.

$1K Capital Is Dangerous

$500 margin at 2x leverage means $1,000 position size. Binance minimum order sizes and maintenance margins eat a disproportionate share at this level. One bad 8h funding period at -0.1% costs you $1 — small in dollar terms but 0.2% of your margin. Over a bear week with negative funding, you could lose 2-5% of your total capital just from reversed funding payments.

Challenger Verdict: USE WITH CAUTION ⚠️ (agrees on core viability but at MAX 2x leverage, not 3x)

⚔️ HEATED

BUILDER: Fair point on the 3x leverage. I'll drop the recommendation to 2x max. But the bear market argument is overblown. The bot EXITS when funding drops below 0.01%. It's not sitting there collecting negative funding. The kill switch is: no position when funding is negative. You're flat during bear markets, not bleeding.

CHALLENGER: The exit logic is the right answer. But funding can flip from positive to negative within a single 8-hour period. If you entered at 0.05% and the market dumps, the next funding period could be -0.1%. You've already collected one positive payment but now you're paying a bigger negative one. The bot needs to be FASTER than 8-hour cycles — it should monitor predicted funding (which Binance publishes) and exit BEFORE the rate flips.

BUILDER: Agreed. Use predicted funding rate, not current. Exit when predicted rate drops below 0.02%. That gives a buffer before it goes negative.

⚖️ DISTILLER

R1 Verdict: USE WITH CAUTION ⚠️

Conditions: Max 2x leverage. Exit on predicted funding < 0.02%. Auto-deleverage at 60% margin loss. Accept 0% returns during bear/sideways markets.

R2 Summary

Builder R2: With fixes applied, revised returns to 3-5%/month (lower ceiling, more honest). Added multi-pair diversification: spread across BTC, ETH, SOL perps. If one pair's funding drops, others may still be positive. Reduces concentration risk.

Challenger R2: Concedes the predicted-funding exit logic is smart. New attack: exchange risk. All capital on Binance. If Binance freezes withdrawals (happened to FTX), 100% loss. Mitigation: split across Binance + Bybit. But at $1K, splitting capital makes positions too small to be meaningful.

Distiller R2: Exchange risk is the remaining 🔴 PERMANENT. Can't be eliminated, only mitigated. At $10K, split Binance/Bybit 60/40. At $1K, accept single-exchange risk as part of the deal.

R2 Verdict: USE WITH CAUTION ⚠️

R3 Summary

Builder R3: With weekly profit extraction, maximum loss is capped at deployed capital + 1 week of profits. Over 6 months, you've extracted enough that even total exchange failure doesn't wipe your cumulative returns.

Challenger R3: Agrees. Weekly extraction is the correct risk management. No new attack vectors. Upgrades verdict.

Distiller R3: Both sides converge. Core strategy is proven, leverage amplification is real, risk management is adequate.

R3 Verdict: USE WITH CAUTION ⚠️

Final Assessment

Kill Metrics

• Funding rates negative across all pairs for 30+ days → shut down, go flat
• Any exchange freezes withdrawals → immediately exit all positions on other exchanges
• Cumulative loss exceeds 15% of starting capital → stop and review
• Margin utilization exceeds 70% on any position → auto-deleverage

🔨 BUILDER

What It Is

Every week, rank the top 50 altcoins by 7-day price momentum (simple % change). Buy the top 5. Sell anything that drops out of top 5. Rebalance weekly. Pure momentum factor — the most documented edge in financial markets.

Why Momentum Works

Momentum is the most studied anomaly in finance. Jegadeesh & Titman (1993) documented it in equities. It's even stronger in crypto because:

• Crypto is dominated by retail (emotional, trend-following buyers)
• No short-selling friction = trends extend longer
• New narratives create prolonged momentum (AI tokens, RWA tokens, memecoins)
• Academic evidence: Grobys et al. (2019) found crypto momentum generates 2.4% weekly excess returns

Economics

The Bull Market Dependency

This strategy ONLY works in bull/trending markets. In bear markets, "top momentum" means "least negative" — you're still holding coins going down, just the ones going down slowest. The Builder is transparent: this is a bull market amplifier, not an all-weather strategy.

Builder Verdict: USE WITH CAUTION ⚠️

🔴 CHALLENGER

The 2022 Backtest Problem

Momentum rotation in crypto during Jan-Nov 2022 (bear market): the top 5 momentum coins each week were losing 5-15% less than the market. You'd still be down 60-70% over the year. The "momentum" just meant you lost less, not that you won.

Turnover Kills Returns at $1K

If 3 of 5 coins rotate weekly, that's 6 trades/week (3 sells + 3 buys). At 0.1% per trade on Binance, that's $1.20/week on $1K. $4.80/month. If monthly returns are $50 (5%), fees eat 10%. At lower return months (2-3%), fees eat 20-30%.

Survivorship Bias in Academic Studies

Grobys et al. (2019) studied 2014-2018 — a massive bull cycle. The "2.4% weekly excess returns" included the 2017 mania where everything went up 1,000%. Applying those numbers to 2022-2023 or any bear market produces catastrophic losses.

Challenger Verdict: USE WITH CAUTION ⚠️ (agrees it works in bull markets, disagrees it's appropriate without a market regime filter)

⚔️ HEATED

BUILDER: The Challenger's bear market criticism is correct. Adding a regime filter: 200-day moving average of BTC. If BTC is above its 200MA, run the strategy. If below, go 100% stablecoin. This is the simplest, most backtested regime filter in existence.

CHALLENGER: Now we're talking. With the 200MA filter, 2022 would have had you in stablecoins from mid-January to late-October. That avoids 80% of the drawdown. But the filter adds whipsaw risk — BTC crosses the 200MA, you buy in, it drops back below, you sell at a loss, it recovers. The 200MA isn't magic.

BUILDER: Add a 5-day confirmation buffer. BTC must be above 200MA for 5 consecutive days before re-entering. Reduces whipsaws by ~60% based on historical backtests.

⚖️ DISTILLER

R1 Verdict: USE WITH CAUTION ⚠️

Conditions: Must include BTC 200MA regime filter with 5-day confirmation. Must go 100% stablecoin when filter is off.

R2 Summary

Builder R2: Added regime filter + 5-day confirmation. Backtested 2020-2025: strategy active ~55% of the time. During active periods, 8-18% monthly returns. During inactive periods, 0% (stablecoins). Blended annual: 40-80% depending on bull cycle length.

Challenger R2: Backtest looks good but it's fitted to a period that included the biggest crypto bull run in history (2020-2021). More conservative estimate: 5-12% during active months, 0% during inactive. Blended annual: 25-45%. Still compelling. New concern: altcoin liquidity at $200/position ($1K capital) — some top-50 altcoins have thin order books, slippage will be worse than modeled.

Distiller R2: Challenger's revised estimate is more honest. Accept 5-12%/month active, 0% inactive. Liquidity concern at $1K is valid but manageable by restricting to top 30 coins by market cap (not top 50).

R2 Verdict: USE WITH CAUTION ⚠️

R3 Summary

Builder R3: Restricted to top 30 by market cap. Added stop-loss: if any single coin drops 15% from entry within the week, sell immediately, don't wait for rebalance. This caps single-position losses.

Challenger R3: Stop-loss is the right addition. No new attacks. This is a clean, well-understood strategy with documented edge. Upgrades to full agreement.

Distiller R3: Unanimous. Simple, documented, automated, low cost. The 200MA filter is the key innovation that makes it viable.

R3 Verdict: USE WITH CAUTION ⚠️

Final Assessment

Kill Metrics

• 3 consecutive active months with < 3% returns → review coin selection criteria
• Max drawdown exceeds 30% in any single month → pause and review
• BTC below 200MA for 6+ consecutive months → strategy hibernating, review capital allocation
• Total fees exceed 25% of gross returns → switch to lower-fee exchange or reduce rebalance frequency to bi-weekly

R1 Summary

Builder: Grid trading on SOL/USDT and DOGE/USDT. Set 20-30 grid levels within a price range. Bot buys at each level on the way down, sells on the way up. Every completed buy-sell cycle earns the grid spread. Works beautifully in sideways/choppy markets (which crypto spends 60% of time in). Pionex offers free grid bots, 0 coding needed. Returns: 3-8%/month in sideways markets. At $1K: $30-80/month. At $10K: $300-800/month.

Challenger: Grid trading was already evaluated in Incubator v1 and got killed in Round 2 for trend risk. A sudden 30% crash means you're holding bags at every grid level with no buyers below. The strategy literally only works when the price stays in your range. If SOL crashes from $150 to $80, your grid bot just DCA'd into a 47% loss. This is the same strategy, same problem.

Distiller: Both correct. Grid trading is profitable in sideways markets and deadly in trending ones. Without a trend filter, it's a coin flip on whether you hit the right market regime.

R1 Verdict: USE WITH CAUTION ⚠️

R2 Summary

Builder R2: Added a regime filter (same 200MA concept). Only run grid bot when BTC volatility is high but direction is neutral (Bollinger Band width > 5%, RSI between 40-60). This isolates the "choppy" regime where grids thrive.

Challenger R2: The filter is more complex than the momentum strategy's simple 200MA. More parameters = more curve fitting = more likely to fail out-of-sample. And the core problem remains: one crash event wipes months of grid profits. Adding filters doesn't fix the asymmetric risk profile. Downgrades to DON'T USE.

Distiller R2: Challenger's curve-fitting concern is valid. The grid strategy requires a more complex regime filter than momentum, which suggests it's less robust. However, the core mechanic works — Pionex has millions of users profiting from grid bots.

R2 Verdict: DON'T USE ❌ (Challenger's downgrade tips the balance)

R3 Summary

Builder R3: Simplified. Drop the complex filter. Instead: run grid bot with a hard stop-loss at 15% below the grid's lower bound. If price drops through the entire grid AND 15% below, liquidate everything and go flat. Maximum loss per deployment: 15-20%. Restart when conditions improve.

Challenger R3: The hard stop-loss is the simplest and best fix. It caps the worst case, which was the original kill shot. Upgrades back to USE WITH CAUTION. However, the stop-loss will trigger often in volatile crypto — expect to get stopped out 2-3 times per quarter, eating into cumulative returns.

Distiller R3: Hard stop-loss resolves the asymmetric risk. Expect net returns of 2-5%/month after accounting for stop-out events. Lower ceiling than momentum rotation but works in different market conditions (sideways vs. trending).

R3 Verdict: USE WITH CAUTION ⚠️

Final Assessment

Kill Metrics

• Stopped out 3 times in a single month → market isn't suitable, pause for 30 days
• Cumulative returns negative after 3 months → strategy isn't working in current market
• Grid never completes a full cycle in 14 days → range is wrong, reset parameters

3-Round Summary

R1 — USE WITH CAUTION: Builder argued that top Solana wallets (tracked via Birdeye, Cielo, GMGN) show 100-500% returns over 30 days. Copy their trades with 5-second delay and position limits. At $1K, copy 3 wallets with $333 each. Potential 10-30%/month.

R1 Challenger: Survivorship bias is catastrophic here. You see the top 10 wallets that made 500%. You don't see the 10,000 wallets that lost everything. The wallets you're copying might be insider wallets that had advance knowledge of token launches. By the time you copy, the insider has already taken profit and you're the exit liquidity. Verdict: USE WITH CAUTION (barely).

R2 — DON'T USE: Challenger found specific examples of "top wallets" that were actually Pump.fun deployers or team wallets. They appear profitable because they created the token, bought early, and dumped on followers. Copy-trading these wallets means you're systematically buying at the top. Builder couldn't refute this.

R3 — DON'T USE: Builder attempted to fix with "only copy wallets with 90+ day track records and 100+ trades." Challenger showed that even filtered wallets have massive drawdown months — the 30-day returns snapshot hides that most top wallets have a 50%+ loss month within every 6-month period. The selection criteria can't filter for this.

Final Verdict: ❌ 1/3 FAIL — DON'T USE

Core problem: you can't distinguish skill from insider advantage from survivorship bias using on-chain data alone. You're always one step behind.

3-Round Summary

R1 — DON'T USE: Builder proposed monitoring Binance open interest, funding rates, and leverage ratios. When leverage is extreme (OI/market cap ratio > 3%), open a position anticipating a cascade. Challenger immediately killed it: you can see leverage is high, but you can't predict WHEN the cascade happens or in WHICH DIRECTION. High leverage means a big move is coming — but you're betting on direction, which is exactly the prediction game that loses long-term. Builder couldn't refute.

R2 — DON'T USE: Builder tried adding "only short when funding > 0.15% AND OI is at 90-day high." Challenger: this happened on November 5, 2024. BTC was at $68K. Funding was 0.2%+. OI was at all-time high. The "cascade" prediction would have shorted BTC. It then pumped to $99K. Specific, dated, devastating counterexample.

R3 — DON'T USE: Builder pivoted to "use it as a hedge trigger, not a standalone strategy." Challenger: if it's just a hedge, it's not generating 5-15%/month — it's an insurance cost. Doesn't meet the brief.

Final Verdict: ❌ 0/3 FAIL — DON'T USE

Core problem: predicting the timing and direction of liquidation cascades is prediction trading, not systematic edge extraction. You're gambling with extra data that doesn't actually help.

BUILD ORDER

1. Altcoin Momentum Rotation (Day 1-2): Simplest to build. Weekly cron job, Binance API, top-30 coin scanner, 200MA filter. Zero cost to run.
2. Leveraged Funding Rate Arb (Day 2-3): Binance API, funding rate monitor, auto-entry/exit, predicted rate tracking. $15-20/month VPS.
3. Grid Trading — $10K only (Day 4): Pionex free grid bot. 30 minutes to set up. Or custom via Binance API in 1 day.

Paper trade all 3 for 14 days before deploying real capital.