brucewedding

If you can make a Polymarket bot work when BTC flips 13+ times $150-400 each way in 3 minutes, you have my respect. I'm just going to pause until things settle down. This has been nuts since last night.

@bolawasfy44 @DextersSolab Yeah, I tried too. Looking at 15 different signals then later using an ML model trained on over 15,000 markets. Still couldn't do better than break even. These markets have a disproportionate share of reversals at T-5 or less.

@DextersSolab you need 99% win rate which is not feasible, i tried it and worked out for a bit but then atopped working risk to reward is so high

This guy makes $1,000 a day and never takes a risk. He found a loophole most traders ignore. Just BTC direction markets. Last 60 seconds. 99% odds only. And $20,700 profit in 1 month is a proof. Here's how it works: Polymarket runs BTC Up/Down markets that resolve every few hours. In final seconds (when the outcome is basically decided), odds on the winning side still haven't fully reached $1. They sit at 98-99 cents. Most traders ignore that last minute completely. He doesn't. He watches BTC price live and waits. 30 seconds left. Direction is clear. No real reversal possible. He drops a large position at 99 cents. Market resolves at $1.00. Done. 1% return. Repeat again. His profile: [polymarket.com/profile/0xaad0…] Sounds boring until you run the numbers. 1% daily on a serious bankroll compounded over 2 months is not boring. It's $21k with almost zero downside. The risk here is basically just a flash crash in the last second. Which happens, but rarely enough that the edge stays massively positive. (You can see small downfalls at his PnL curve). I've talked about last-second market entries before. This is the cleanest version of that strategy i've seen executed consistently. No drama. No major losses. Just quiet daily collection. Safest $1,000 a day on Polymarket right now.

1. You can't snipe from the UI. You need a bot with the chainlink RTDS feed. 2. The market was swinging $400 each way last night and $100 during the market you traded. Crazy to try to snipe in those conditions 3. Sniping is always a high risk, low reward activity. You have to be right about 80X more often than you're wrong. It's not remotely as easy as it seems. Markets do flip last second. 4. You shouldn't be gambling with money you need to feed your family. That's just dumb. 5. Polymarket resolves on Chainlink, not the UI and the rules clearly state that on each market. 6. You're not going to get this money back. Expensive lesson.

@devjoshstevens @mustafap0ly That was supposed to be 18ms, not 28ms. Even better.

@devjoshstevens @mustafap0ly I'm a guy that gives credit when deserved. Not sure what changed but the API is kicking ass lately. I'm averaging 28ms order placement and cancels. Thanks.

@Gustafssonkotte No show it sorted by date instead of profit. Way too many gaps in those markets to think you're not full of shit.

1/7 Built a Polymarket trading bot over 3 months. Here are the biggest mistakes that cost me real money > Went from v1 to v61. Every version fixed something painful. --- 2/7 Stop Loss killed more money than it saved. > Binary markets need room to breathe - fluctuations are normal. > Stop Loss was cutting positions on random noise and locking in losses right before the market flipped. > Removed it in v61. Immediately better. --- 3/7 Martingale + Stop Loss = a loss cascade. > Seemed logical: lost $5 -> bet $8, lost again -> bet $10. > In practice: a losing streak plus early exits = a hole in your balance in a single day. > Killed it. For good. --- 4/7 Smart Exit without Force Exit is a trap. > Token hits 90c (+75% profit), but the bot was waiting for a "BTC reversal" signal. > Market closes, token drops, profit gone. > Fix: hard Force Exit at 85c. No conditions, no waiting. --- 5/7 Blocking the 5:30-10:30 PM ET window felt safe. It wasn't. > NYSE open = sharp spikes = bad signals. Made sense to block it. > But the full block was also killing clean entries at 8-10:30 PM. > Had to split the zone into segments with different edge/move thresholds. --- 6/7 The Gamma API lies about market start time. > Start price ("price to beat") is the core input for every signal. > Gamma was returning stale data. Had to pull prices directly from Chainlink on-chain on Polygon. > That's its own adventure - polling a smart contract every 2 seconds at 2 AM. --- 7/7 The real lesson: don't overcomplicate what works. > v1: complex system, 10 indicators -> -$200/day > v61: "buy the expensive token for $5, exit at +30%" -> consistently green > Simpler logic = fewer failure points. > The bot runs 96 intervals a day. Every mistake shows up fast.

Their claims don't hold up on our larger dataset of 11,000 markets. Fade the rip: Their claim: 0.30-0.40% pump -> 60.6% DN win rate (274 trades) Our data: 52.8% DN win rate (161 trades). Barely above coin flip. The 0.35-0.40% bucket shows 57.3% on 75 trades -- interesting but thin sample. Buy the dip: Their claim: doesn't work Our data: -0.30 to -0.35% dip -> 64.2% UP win (109 trades). Actually works better than their fade! European hours: Their claim: 58% in Europe Our data: 53.9% in Europe. Nothing special. The EV looks positive but it's misleading. The avg DN ask at open is $0.30 -- that means both sides are near $0.50, the market hasn't leaned yet. You're buying at $0.30 and winning 53% -- that's positive EV only because the entry price is so low. It's the same "buy before the lean" dynamic, not a fade signal. Bottom line: On 11.5K sessions (vs their 9K), the fade signal is noise. 52.8% on 161 trades is well within random variance. Their "sweet spot" at 0.30-0.40% is likely overfitted to their smaller dataset.

This strategy is printing on Polymarket right now. Fade every BTC pump. Bet DOWN. 9,075 BTC 5-minute markets. Binance 1-second price data. We tested every "buy the dip" and "fade the rip" variation. BTC dropped, bet UP (bounce): Dropped 0.1% in 5min: 48.7%. Coin flip. Dropped 0.2% in 15min: 50.6%. Coin flip. Dropped 0.5% in 30min: 54.6%. Barely there. Dropped 1.0% in 60min: 45.5%. Actually worse. Buying the dip doesn't work. At all. But fading the rip does. BTC pumped in the last 15 minutes. Bet DOWN: Pumped 0.1%: 52.5%. Pumped 0.2%: 56.7%. Pumped 0.3%: 62.5%. 168 trades. Pumped 0.5%: 69.8%. 53 trades. And it only works in a specific range. We zoomed into the previous 5-minute market's BTC move: Previous move 0.00-0.20%: reversal 49.5%. Coin flip. Previous move 0.30-0.40%: 60.6%. 274 trades. Previous move 0.40-0.50%: 50.0%. Gone. Below 0.30%, not enough overshoot. Above 0.40%, it's real momentum. The sweet spot is a narrow 0.10% window where BTC consistently overshoots just enough for a bounce. With real opening token prices at $0.546: EV = +$1.86 per $10 trade. Positive even after the market prices in a slight reversal bias. During European hours (8-16h UTC): 58.0% reversal on 231 trades. US and Asia: 52%. Why does fading work when dip-buying doesn't? Short-term pumps overshoot more than dips. Aggressive buyers push price past fair value. The next 5 minutes correct it. If you want to backtest your own strategies, polybacktest has subsecond historical data going back a month

@devjoshstevens @mustafap0ly Guys, for the love of God, fix the goddamned websockets. I can't even go a single 5m market on XRP without a disconnect. Same code works fine, on the same endpoint for BTC

@MarcusAkvilla @Abombination81 Dude, from websocket -> json parsing -> filters -> order creation -> signature -> post order, my bot is < 200 microseconds. IOW, 1500X faster than yours. Python is your problem. I run C++. Rust works too.

@Abombination81 I don't understand how your execution speed gets faster. I have a stack of: VPS Ireland, Node js, Python, FAK warrant.Timing: signal→preflight: 1ms preflight→order: 5 ms ← signature postOrder: 339ms ← Polymarket server total: 448ms

Looking to start a discord group with other advanced bot makers. We can share, improve and collaborate on bots that don't have liquidity limiting issues. I also have a large, audacious plan to link all the prediction markets and what sports books we can, together for a multi market arb monstrosity. If you're making money on more than one bot, and are interested, pm me please.

@devjoshstevens @Polymarket Thanks. Don't forget to beef up the websockets which have been shit lately. No data and random disconnects. Worse on XRP and SOL, than BTC. My code hasn't changed in 2 months and this is new in the past few weeks.

@brucedwedding @Polymarket Yeah not acceptable sorry for that we are going to spend time optimising the CLOB to be stable and fast all the time.

Excited to join @Polymarket as VP of Engineering. I haven’t seen a hungrier team in my career, the talent density here is insane. We’re building the future of how people understand the world. What should we build next? What’s broken? What’s missing? Drop it below. P.s we are hiring world class people in NYC DMs open

@direkturcrypto @fatedotsol Let me save you some time and money. If you're trying to get profitable pairs at the same time, you can't.

@fatedotsol yes, buy UP & DOWN below ¢100

Me: I'm going to build HFT on Polymarket. Also me after 1 month: 🤣💀 Current status: > RPC lagging > API lagging > Orders duplicating > One side filled, other side: ghost 👻 > Partial fills everywhere > Entry too early > Entry too late > Re-entry broken > Strategy still not profitable And my capital? Also lagging. To zero. 😭 The dream: HFT = fast money The reality: HFT is just finding new and creative ways to lose money faster than a normal person. Almost 1 month in. Wallet getting thinner. Confidence getting thinner. Hair getting thinner. But the bot? Still running. Still losing. Still trying.

@thiojoe When your computer is powered off, everything in RAM (your entire OS and state) is gone.

Anyone else think it's weird that computers take even multiple seconds to boot up? I feel like they should take milliseconds

@rcom1337 Please nobody try this with money you can't afford to lose, because you will.

Literally free money on 5-min BTC markets. How does it work? It might look very obvious, but all you need is just to enter in the last seconds before the outcome happens. Let me explain: > open BTC/USDT/BINANCE on TradingView > wait for the 30 seconds on the 5-minute market > if the gap is far enough -> entry > if the gap is so close <$35-50 -> skip But be aware, sometimes there is a kind of glitch when everything reverts in the last seconds, that's exactly why you should look at the TradingView chart super carefully. $100 a day, not that hard, are you in?

@rcom1337 @nodefounder LOL! You just said you're a loser. 90% isn't good enough to make this strategy profitable.

@nodefounder TradingView works fine for me in 90% cases, of you can do it your own tactic

@TradePolyBot Say you don't understand prediction markets without saying it.

61 wins out of 62 trades on Polymarket. here is the exact strategy: 15m all markets entry after 870s (30s left in the cycle) enter at 90¢. SL at 49¢. slippage 10% this is not theory. someone in the PolyBot community is running this right now and sharing the proof.

At small scale, you have to implement active balancing strategy. If you get too heavy on one side, you stop buying until you pair up the light side. And yes, you'll find the market makers are more than happy to sell you the losing side. What I do is buy the high side first, then pair up with the low side, which is easy to do. I seldom wind up long on the losing side. Then I have an active "end rebalance" stage which makes every effort to get balanced before the end. You have to do that math and realize, 20 pairs at $1.10 costs you $2.00. 20 losing pairs at .50c costs you $10. Better to pair up and eat the $2.

@Abombination81 Arb bot, successful when round gets fully matched but a lot of time will end with imbalance with the winnings side always less shares than losing side. And it’s a slow drainer at that point. Run on AWS Ireland, rust + python. How to fix? lol

Lets Fix Your PolyMarket Bot! You have a bot, its not profitable or not working right, this is your spot to ask whatever questions you need! Read the rules! Some rules 1. do not share your edge. An edge is the thing you have that makes you win more than just a coin flip. If you don't know if you have one, you do not. 2. Describe your bot. What markets (weather, 5m BTC, all 15m crypto etc). A quick description of the issue (wont hedge, too slow, no fills etc). The more info, the easier it is to fix. 3. Tell me your tech stack. Is this python, rust, combo? Are you on a vpn, home computer, deployed to a virtual machine? 4. Keep it specific to making your bot work, we have other threads for other things.

I love when my trading bot goes nuts and buys 10 shares at a time every 300ms rather then once. I stopped it around 180 shares and fot lucky. That side won.

Live logs from an Amsterdam server: [20:12:06.775] [btc] Position updated: UP=6, DOWN=0 (avg UP $0.29, avg DOWN $0.00, ( $0.29 ) unpaired UP 6 @ $0.29) [20:12:18.662] [btc] Batch cancel succeeded [13839ms] [20:03:10.007] [btc] SOLD DOWN 6 shares @ $0.63 (avg cost $0.52, P&L +$0.66) [20:03:10.007] [btc] Position after sell: UP=0, DOWN=0 (avg UP $0.00, avg DOWN $0.00) [20:03:26.307] [btc] Batch cancel succeeded [18049ms] [20:01:57.812] [btc] TRADE FILL (MAKER): DOWN 6.0 shares @ $0.5200 [20:01:57.812] [btc] Position updated: UP=0, DOWN=6 (avg UP $0.00, avg DOWN $0.52, ( $0.52 ) unpaired DOWN 6 @ $0.52) [20:02:06.120] [btc] Batch cancel succeeded [11600ms]

@devjoshstevens @Polymarket The CLOB API has periods where it can literally take 10-20s to execute an order. I've had to add defensive protection for it in my bot because I lose a shit ton of money when it happens. Normal execution times are 30-200ms, which works. 14 seconds does not.

@Emmett31539642 @iAnonymous3000

then why was it hacked and encryption beaten? Open source does not mean what you think it does. any hacker can take the source code and defeat the encryption just by reverse engineering the source code. the encryption protocols are within the source code anyways. something signal totally fucked up on. APP encryption can be broken and defeated. WHY? because it doesn't have keys. this is a crucial aspect of encryption. if you don't have keys, encryption is useless. Signal is NOT a private app, it's commercial and the public at large can use it. PLEASE do some research before you post and stop being like @unixpill

You don’t need to trust Signal. That’s the whole point. The protocol is open source. The clients are open source. The server is open source. Independent cryptographers have audited the Signal Protocol repeatedly, and it remains the gold standard for end-to-end encryption. Signal also supports reproducible builds - meaning you can compile the source code yourself and verify it produces the exact same binary distributed on app stores. Trust is for closed systems.