Articles
Analysis, strategy breakdowns, and weekly performance reviews from the CoinClaw competition.
-
Three Experiments, Three Failures — What CoinClaw Learned From Strategies That Don't Work
—
CoinClaw ran three strategy experiments this week: BTC Grid with regime filter, SOL Breakout with regime filter, and BTC Mean Reversion. All three failed Gate 1. Here's what the data says about which strategies actually work.
-
The V3.5 Paradox — A Bot That Failed Validation Is Up 5.59% on Real Money
—
V3.5 Grid was deployed before the validation framework existed. It failed Gate 1 with p=0.938. It's been running for 21 days with real money and is up $33.93. What's going on?
-
How a $80,000 Ghost Price Froze V3.6 for Three Weeks — And How the Fix Works
—
V3.6 Fear & Greed bot stopped trading because its last_price was stuck at $80,000 from a developer's machine while BTC was at $68,600. Here's what happened and how stale price detection fixed it.
-
Best Crypto Trading Bots in 2026: Honest Comparison With Real P&L Data
—
We compare 3Commas, Cryptohopper, Pionex, HaasOnline, and our own CoinClaw bots — with real performance data most platforms won't publish. Updated April 2026.
-
BotVersusBot Scoreboard — April 10, 2026
—
Live scoreboard for the BotVersusBot crypto bot competition. V3.7 Scalper up +$8.22 P&L. V3.8 ETH Grid live at $1,558 equity in bull regime. BTC at $71,926. Updated April 10, 2026.
-
BotVersusBot Scoreboard — April 9, 2026
—
Live scoreboard for the BotVersusBot crypto bot competition. V3.7 Scalper up +$7.60 P&L. V3.8 ETH Grid now live at $1,793 equity. V3.5 Grid intentionally paused. Updated April 9, 2026.
-
The 6th Experiment — Why More Trades Didn't Save ETH Grid on the 1-Hour Timeframe
—
CoinClaw's 6th strategy experiment tested ETH Grid on a 1-hour timeframe. It passed Gate 1 with p=0.002 and 1,357 trades — then failed Gate 2 with a WFE of 0.021. More data points don't fix overfitting.
-
BotVersusBot Scoreboard — April 8, 2026
—
Live scoreboard for the BotVersusBot crypto bot competition. V3.7 Scalper leads with +$3.87 P&L. V3.5 Grid intentionally paused. V3.6 FG recovered and running in extreme fear. V3.8 ETH Grid paper trading at +$18.31. Updated April 8, 2026.
-
The Strategy Research Roundup — 5 Experiments, 1 Winner, and What CoinClaw Learned About Finding a Real Edge
—
CoinClaw's validation framework tested 5 strategy experiments across BTC, ETH, and SOL. Only ETH Grid Config B passed all three gates. Here's the full data, the analysis, and the V3.5 paradox that challenges everything.
-
How a Test Suite Took Down a Live Trading Bot for 9 Hours
—
A CoinClaw test wrote mock prices to the live market data cache. V3.6 read $80,000 BTC when the real price was $68,125 — and shut itself down. Here's what happened and how it got fixed.
-
V3.8 Gets Dynamic Capital Allocation — Why Fixed Grid Sizing Had to Go
—
CoinClaw's most validated bot just learned to size positions based on available capital. Here's why fixed allocation was a problem and how dynamic sizing works.
-
The ETH/USDC Pivot — When the Most Validated Bot in CoinClaw History Hit an Exchange Constraint
—
V3.8 ETH Grid passed all three validation gates with the strongest results ever recorded. Then it couldn't trade. Here's the story of the ETH/USDT to ETH/USDC pivot and what it teaches about deploying validated strategies to live exchanges.
-
Why BTC Trend Is Down $97 — What Happens When a Trend-Following Bot Meets a Ranging Market
—
BTC Trend is the worst-performing bot in the competition at -$97.18 on $7,000 capital. Here's what went wrong, why trend-following strategies struggle in ranging markets, and what the data tells us.
-
The First 30 Hours — What Happens When You Run 3 Live Trading Bots on One Machine
—
CoinClaw migrated 3 live trading bots to a new server and ran them for 30 hours straight. V3.5 is down 2.46%. V3.7 is up 1.28%. V3.6 was frozen by a ghost price. Here's the full performance story with real numbers.
-
5 Experiments, 1 Winner — Why Finding a Validated Trading Strategy Is Harder Than You Think
—
CoinClaw ran 5 strategy experiments across BTC, ETH, and SOL. Only 1 passed all validation gates. Here's what failed, why it failed, and what the results tell us about systematic trading.
-
How Grid Trading Bots Work — The Strategy Behind V3.5, V3.7, and V3.8
—
Grid trading is the most common strategy in the CoinClaw bot competition. Here's how it works, why three different bots use variations of it, and what makes each one different.
-
Behind the Scenes — What It Takes to Keep Live Trading Bots Running (V3.8 Deployment + ccxt Fix)
—
BotVersusBot goes behind the scenes: V3.8 ETH Grid passes all 3 validation gates and earns live approval. Then a ccxt 3.1.60 compatibility bug takes down all 3 live bots for 4+ hours. How the CoinClaw team diagnosed the issue, built the fix, and what it reveals about the operational discipline required to run live trading bots safely.
-
Market Mood Shift — V3.8 Flips to Range Mode While Fear & Greed Hits 11. Here's What Every Bot Is Doing.
—
The Fear & Greed Index just hit 11 (extreme fear) while BTC holds $68K. V3.8's regime filter flipped from bull to range. V3.7 Scalper quietly grinding +$0.59. V3.5 opened its first position in hours. A real-time look at how 4 different strategies respond to the same market.
-
Mid-Week Update — V3.5 and V3.7 Running Clean. V3.6 Stuck on a Price Ghost. V3.8 Paper Watching ETH in Bull Mode.
—
CoinClaw mid-week bot update: V3.5 Grid and V3.7 Scalper recovered after the ccxt fix. V3.6 Fear & Greed bot stuck on stale price data. V3.8 ETH Grid paper bot tracking bull regime. Current standings and what's next.
-
BotVersusBot Scoreboard — April 7, 2026
—
Live scoreboard for the BotVersusBot crypto bot competition. All 3 live bots blocked by ccxt 3.1.60 fix (pending merge). V3.8 ETH Grid approved for live deployment. Paper bots active: BTC Grid Range +$0.91, V3.8 ETH Grid -$0.48. Updated April 7, 2026.
-
The Three Gates — How CoinClaw Decides Which Bots Deserve Real Money
—
CoinClaw uses a three-gate validation framework to decide which trading bots go live. Gate 1 tests for statistical edge. Gate 2 tests for walk-forward efficiency. Gate 3 tests for regime robustness. Here's how each gate works and why most bots fail.
-
V3.8 ETH Grid Goes Live — The First Validated Bot to Graduate from Paper to Real Money
—
V3.8 ETH/USDT Grid is the first CoinClaw bot to pass all three validation gates and get approved for live trading. Here's what that means, how it got here, and what happens next.
-
What Is BotVersusBot? A Beginner's Guide to the CoinClaw Trading Bot Competition
—
BotVersusBot tracks a live competition between algorithmic trading bots — some trading real money, some on paper. Here's what the competition is, who the contenders are, and how to follow along.
-
BotVersusBot Scoreboard — April 6, 2026
—
Live scoreboard for the BotVersusBot crypto bot competition. Krypto runs 3 live bots ($1,607 capital) and 2 paper bots (+$15,490 combined). Key runs 2 active paper bots (+$29.92 BTC Grid Range, -$59.79 BTC Trend). 2 validated strategies remain paper-only. Updated April 6, 2026.
-
Sprint 17 — V3.8 Approved. Bots Come Home. The Live Migration Begins.
—
BotVersusBot Sprint 17: V3.8 ETH/USDT Grid passes all 3 validation gates and gets live bot approval (PR #1078). CoinClaw bots migrate from Delmar's decommissioned machine to the devdsk — and immediately hit Binance API key errors. The Underwater Position Tracker ships (PR #1080). Bot migration merges (PR #1077). The competition enters a new chapter: infrastructure week.
-
Week 16 — BTC Holds $67K. Key's Trend Bot Posts First Positive Day. Grid Bots Keep Grinding.
—
BotVersusBot Week 16: BTC holds near $67,115 as Fear & Greed inches from 12 toward recovery. Key's BTC Trend bot posts its first positive daily P&L (+$9.63) after twelve consecutive losses. Key's BTC Grid Range passes all 3 validation gates with 39 trades at 100% win rate. Krypto's V3.6 gate remains blocked at F&G=12. The live competition widens as Krypto runs three profitable live bots while Key awaits deployment approval.
-
Week 15 — BTC Slips to $67,115. F&G Climbs to 12. Key's Trend Bot Closes All Positions and Goes Flat.
—
BotVersusBot Week 15: BTC eases from $68,127 to $67,115 as Fear & Greed inches up from 8 to 12 — still deep in Extreme Fear. Key's BTC Trend bot closed its three re-entry positions on April 1 via EMA_CROSS_DOWN at $68,007, recording three small losses to extend the run to twelve straight. Key is now flat with zero open positions. Krypto's V3.6 picked up 8 more closed trades. Both competitors await a sentiment shift that hasn't arrived.
-
Week 14 — BTC Bounces to $68,127. Fear & Greed Hits a New Low of 8. Key's Trend Bot Loses Nine Straight Before Re-Entering.
—
BotVersusBot Week 14: BTC recovers from $65,929 to $68,127 — but Fear & Greed drops to 8, a new competition low. Key's BTC Trend bot recorded three more losses (MACD_TURNED_NEGATIVE on March 30) for a total of nine straight, then opened three new positions on March 31. The April 1 live gate has fired. Krypto's V3.5 lower-bound threat has eased.
-
Week 13 — BTC Falls to $65,929. Krypto's V3.7 Scalper Goes Live. Key's Trend Bot Records Six Consecutive Losses.
—
BotVersusBot Week 13: BTC slides further to $65,929 as F&G hits 9 — deepest Extreme Fear of the competition. Krypto's V3.7 Narrow Grid Scalper launched live last night. Key's BTC Trend bot now 0/6 in its current session. The April 1 live gate is tomorrow.
-
Week 12 — BTC Drops to $66,725. Krypto's Grid Has $915 of Runway. Key Takes Three Stop-Losses in a Row.
—
BotVersusBot Week 12: BTC slides 5.6% to $66,725 as Extreme Fear enters its second consecutive week. Krypto's V3.5 grid is $915 from its lower bound — and V3.6 is going live with $1,000 real USDC. Key's BTC Trend bot took three consecutive stop-losses.
-
Week 11 — Krypto Holds +$18,834 as Market Stalls. Key Files 5 Strategy Specs. BTC at $70.7K, F&G Barely Crawls Off Extreme Fear.
—
BotVersusBot Week 11 update: Krypto's V3.5 paper bot frozen at +$18,010 pending relaunch. Key completes brutal honest audit of 4 bots — 0 edges proven, 5 specs filed. BTC recovers to $70,682, F&G at 14.
-
BVB Relaunch — Week 10: The Competition Returns. Krypto Leads +$18,834. Key Starts Fresh. BTC at $68.6K in Extreme Fear.
—
The BotVersusBot competition is back after a one-week infrastructure pause. Krypto enters the relaunch with +$18,834 on paper — V3.5's 2,481 trade win streak intact, V3.6 carrying 10 underwater positions. Key relaunches with $2.00 across three bots on three assets. BTC sits at $68,671 with Fear & Greed at 10 for the seventh straight day. Week 10 is the first real test of both agents' strategies against a sustained bear market.
-
Week 9 — The Scalper Arrives: V3.7 Goes Live, BTC Bleeds to $68.6K, F&G Sets Competition Low
—
Week 9 brought two developments the competition had been building toward: a new bot going live, and the Fear & Greed Index hitting its lowest reading since day one. V3.7 Narrow Grid Scalper launched on Saturday evening, ran 63 cycles in its first 15 hours, and banked its first two trades while V3.5 continued harvesting below its entry range. BTC is now down $6.5K from the Week 7 peak.
-
Week 8 — The Gate That Lasted Three Days: F&G Collapses 28→11, BTC Sheds $4.4K, Live Bot Holds the Line
—
Week 8 opened with the competition's best setup in eight weeks — Fear & Greed above the gate threshold, BTC recovering, live bot freshly recentered. It ended with F&G back in Extreme Fear (11), BTC at $70.6K, and the live bot holding five underwater positions while the paper fleet sat idle. The gate V3.6 waited seven weeks to open lasted just three days.
-
Week 7 — The Gate Finally Opened: F&G Hits 28, V3.6 Condition Met, Live Bot Recalibrated
—
Week 7 is the week V3.6's Fear & Greed gate was finally breached — sentiment recovering from 15 to 28 in three days, clearing the ≥20 threshold for the first time in the competition. The live V3.5 bot was reset, recentered, and is actively cycling at $74K–$75K. Key's BTC Trend bot executes its first-ever trades. But with crons paused by Delmar directive (March 16), the gate opened to a competition in standby mode.
-
Week 6 — Real Money on the Table: V3.5 Goes Live as BTC's 22% Surge Tests Every Bot's Limits
—
Week 6 marks a competition milestone: Krypto's V3.5 Grid goes live with real money (Delmar override, March 11). Then BTC does something extraordinary — a +15.3% overnight surge from $71,949 to $82,975, followed by a near-full reversal back to $71,561. The grid escapes its range. The live bot battles a stale data cache. Fear & Greed stays stubbornly below 20. Key's bots remain offline. This is the wildest week of the competition yet.
-
Week 5 — BTC Claws Back to $70K: Krypto's Best Night Yet, V3.6 Still Waiting for Fear to End
—
Week 5 opens with BTC recovering to $69,613 and Krypto's grid bot recording its single biggest overnight gain of the competition: +$1,239.17 in one night. V3.6 remains gated as F&G crawls from 8 to 15. Key's bots are fighting infrastructure fires. Here's the full picture.
-
Week 4 — The Fear & Greed Gate Opens: How CoinClaw Bots Navigated F&G=22 and $71K BTC
—
Week 4 final: V3.5 Grid closed at +$1,798 (+179.8%) across a full 7-day window. V3.6 Fear & Greed finished $0.00 — F&G=8 at week end with gates locked throughout. Complete Week 4 data including the rate-limit bypass finding, the backtesting Gate 1 fail, and what the divergence means for Week 5.
-
Fear & Greed Says 10. BTC Says $71K. Week 3 Opens With the Competition's Most Interesting Divergence.
—
Week 3 of the Bot vs Bot competition kicks off with a split: BTC rallied +6.5% in a single day to $71K, while the Fear & Greed Index hit 10 (Extreme Fear). Key's bots traded into the breakout. Krypto's gates stayed locked. Here's exactly what happened.
-
BTC at $63K and Falling: How Two AI Trading Bots Survived Their First Market Crash
—
Week 2 of the Bot vs Bot competition. BTC dropped to $63,462, Fear & Greed hit 11 (Extreme Fear), and both AI trading bots faced their first real defensive test. Here's exactly what happened — numbers, failures, and one real-time fix.
-
Two AI Bots, One Competition: What We've Learnt in 8 Days of Algorithmic Crypto Trading
—
We built two AI trading bots — Krypto and Key — gave them $10,000 each and told them to compete. This is an honest account of what happened in the first eight days: the V1 -17% failure, six signal gates that blocked every entry during February's downturn, and Key's live bot data. Real numbers, no filter.