(Jan 22, 2026 04:21 PM)Omniscient Wrote:  

I didn't even think of this before seeing that post. Who's going to make an AI bot to automate this and make a small fortune?

(Jan 22, 2026 06:22 PM)SUI-3183372 Wrote:  The first involves a total reliance on speed. We're on bare metal servers within the same datacenter as Polymarket in Europe, we're tapped into Chainlink's data streams, and we only look at the 15-minute and 30-minute crypto markets. As we detect large enough shifts in price through chainlink's data streams, we attempt to beat the rest of the market to react to these changes and bet on them on polymarket. Once the rest of the market reacts, we withdraw our bet and reap the 1-5% rewards (sometimes its larger than this in extreme cases). In all cases, we're looking to enter and exit within about 30 seconds, we are not playing these markets to resolution.

(Jan 22, 2026 06:22 PM)SUI-3183372 Wrote:  Third is arbitrage between kalshi and polymarket, with the intent of it being more or less what you displayed in the OP, but it's much cruder in its current form. This is our newest addition and is still be working on, but it should be a bit safer than approaches 1 and 2.

(Jan 23, 2026 11:34 AM)Kify Wrote:  I have been wanting to do this for a while but refrained from it because I always thought that the competition would just kill me. Looking at the JaneStreetIndia bot for example that made 650k while it was still active just reiterated the idea that "the big boys are here" and "companies like WinterMute are already on this and will beat me everytime". If you don't mind me asking, what is the winrate of the bot and average bet size?

(Jan 23, 2026 11:34 AM)Kify Wrote:  Considering Kalshi is not natively onchain but Poly is, this would not be a guaranteed arbitrage, correct? Like a leg could still fail to go through as it's not an atomic bet. Kalshi since recently does have tokenized predictions on Solana but then again Polymarket is on Polygon.

Hella cool man, good luck with the bot, wish I went through with building mine haha, I might just try to make one for a bit more of a niche (and thus, way less liquid unfortunately) Poly category I had my eye on for a while.

(Jan 23, 2026 01:37 PM)◔ßorder╙ineExtrⒶ Wrote:  In layman's terms, what is the difference between High-frequency trading (not speaking about the HF group, lol) and arbitrage? Or just the platform spread invents a whole new category.

(Jan 23, 2026 01:42 PM)SUI-3183372 Wrote:  I don't know how you define high frequency trading, but I'll explain how we approach arb here and let you decide how it fits in with your understanding.

Look for events that exist on both venues, where buying yes on one venue and no on the other results in a combined cost of less than $1.00.

For example, a yes bid at 0.46 on Polymarket and a no bid at 0.49 on Kalshi gives you an all-in cost of 0.95. At resolution, one side pays out $1 and the other pays $0, giving you a $0.05 edge before fees. In Omni's OP, it's $0.89 vs $0.07 = $0.96 all in, setting you up to earn ~4% in value on your initial capital.

(Jan 22, 2026 06:22 PM)SUI-3183372 Wrote:  About 3 months ago someone on X exposed how they're making about $25k per month doing this, and it was piss easy how they were approaching it. This motivated me @Xbox Live@ and Santahat to vibe code up a similar tool. Our tool has three ways to approach prediction markets.

The first involves a total reliance on speed. We're on bare metal servers within the same datacenter as Polymarket in Europe, we're tapped into Chainlink's data streams, and we only look at the 15-minute and 30-minute crypto markets. As we detect large enough shifts in price through chainlink's data streams, we attempt to beat the rest of the market to react to these changes and bet on them on polymarket. Once the rest of the market reacts, we withdraw our bet and reap the 1-5% rewards (sometimes its larger than this in extreme cases). In all cases, we're looking to enter and exit within about 30 seconds, we are not playing these markets to resolution.

The second also relies on speed, but to a lesser degree. We watch these same 15-minute and 30-minute markets as they approach their closing times. As we're seconds away from close, and a outcome seems close enough to certain, but there's still a large enough divergence on polymarket, we'll throw funds at them to capture the 0.5-2% difference between seconds before close and full resolution.

Third is arbitrage between kalshi and polymarket, with the intent of it being more or less what you displayed in the OP, but it's much cruder in its current form. This is our newest addition and is still be working on, but it should be a bit safer than approaches 1 and 2.

We're not earning a killing on this, primarily because we haven't dialed it in perfectly yet so scaling would be too risky, but it's cool seeing this tool winning dozens of times per day, bringing in $0.50 to $5.00 or so per win. Doesn't seem like a lot, but theoretically we can scale this up to be about 575 wins per day, assuming we're perfect, just on method #2. From there, scale up the amount that we bet as well, and we could start bringing in some decent money. This is a hot topic in the space though, I'd expect this to become increasingly saturated over 2026.

(Jan 25, 2026 07:14 PM)d3t0x Wrote:  I vibe coded one over xmas for dayyys. Tried multiple strategies. I was able to make 2-3% profits every 15m window for 2/3/4 hours straight and then one bad trade would wipe out all my profits haha - Tried all sorts of ways to reduce the risk, detect and exit those bad trades early. I even put together a jupyter notebook and did some data analysis on some of the successful bots vs mine but couldn't figure it in the end. Might circle back to it some day. The kalshi x polymarket strat seems interesting! that gives me a new idea.

Analysis of https://polymarket.com/@gabagool22 trades over 15m window