Replying to @elitebrudda @MEXC
MMs should be providing liquidity around the index/fair price
but someone found a way to pump/dump the perps while the underlying stock market is closed and MMs are adjusting their quotes, allowing them to open large shorts at inflated prices
then the price returns toward the index, they close their shorts into MM and repeat. whole cycle appears to be automated and takes just seconds
MMs are taking the losses. and as far as I know, MEXC has its own MM, so it could be their own money on the line
Rossium retweeted
Good places to look for +EV:
1. The idea is hard to backtest
2. The data is hard to acquire
3. A known input can be transformed into something novel or hard to compute.
5. The idea is known, but has only been applied to one specific area or time horizon.
1/n
Rossium retweeted
"Obsessing over your competition is the fastest way to build a mediocre version of their product." - Author Unknown
I built a bot to trade faster than any human
> found markets to bet which word Trump says in speech
> used Deepgram Nova-3 to transcribe live speeches in <0.2s
> matched words with markets
> instantly sweep all orders under 99¢
> exit at $1.00 for profit
> in the right hands, generates 6-figures/month
All done in 2 hours 34 minutes using Perplexity, Opus 4.6, and a bit of brain
None of my CS knowledge was used except deploying it on a server
> be anon
> studies probability theory since 14
> gets into MIT
> recruited by Jane Street
> 2 years, $350k/yr, 80hr weeks
> discovers people bet on 15-min BTC markets
> realizes he’s been training for this his entire life
> builds Poisson + Kelly model in one weekend
> loads $389
> made more in 30 days than 2 years at Jane Street
> quits and goes full-time building quant systems
> now makes 7 figures/month from a laptop in St. Barts, trading in his underwear
never before in human history has it been so easy to escape the permanent underclass
you can literally become quant by reading this article, using AI and being locked in for a 6 months
- but most of you will not
ran simulations on 2,672 USGS earthquake data since 1964
tested model a week ago and it outperformed market. told me to buy "NO" on 0 earthquakes and it won. 124% ROI
this time model says same market is overpriced again
data: 2,672 M6.5+ quakes from USGS (1964-2026)
model: Poisson(λ=0.824/week)
output: market prices "0 quakes" at 56¢ - our model says 44%
buckets 1, 2, 3 are all underpriced
P(0) = 43.9% vs 56% > SELL
P(1) = 36.1% vs 31% > BUY (+5.1% edge)
P(2) = 14.9% vs 11% > BUY (+3.9% edge)
P(3) = 4.1% vs 2% > BUY (+2.1% edge)
learn about quant methods, use AI, escape the permanent underclass
🚨 Susquehanna is hiring prediction market traders at $200k+ to do one thing: build models that detect mispricings on Polymarket and Kalshi
They already have 60 traders doing this.
Goldman Sachs is exploring the space. Fed just said Kalshi is useful for forecasting.
this is the same thing I've been posting about:
> markets mispriced
> cross-platform arbs sitting untouched for days
> prices significantly delayed vs real-time data
Wall Street is building entire desks around prediction markets
this space is about to get very competitive very fast
the window for retail edge is closing
if you're not paying attention to prediction markets yet, you're late
I analyzed 11 past Apple event transcripts for Polymarket mispricings
This is what I found:
'Apple Watch' is priced at 65% YES right now. BUT
> there is no Watch launching
> series 11 came out 5 months ago
> zero rumors
> historically mentioned 0 times at non-Watch events
The 65% comes entirely from actual Watch launch keynotes.
Apple Watch NO at 35¢ is the cleanest trade on the entire board
Full analysis on all 20+ markets dropping soon.
money glitch has been sitting on Polymarket for days and nobody noticed
Let's take a look at 'Will the US confirm aliens exist before 2027?' market
Polymarket YES: 15¢
Kalshi implied YES: 22¢
here's what's funny: Kalshi requires White House verification to resolve YES. Polymarket accepts any federal agency + credible reporting.
stricter criteria should mean cheaper YES on Kalshi, not more expensive
buy YES on Poly at 15¢, buy NO on Kalshi at 78¢ = locked in 7¢ spread regardless of outcome
you don't even need to hold until resolution
wait for the prices to converge and exit once the gap tightens
trades like this happen all the time
Earthquake markets are mispriced on Polymarket
Lets look at 'How many 7.0 or above earthquakes in 2026?' on @Polymarket
0 quakes in the first 50 days
retail is anchoring to a slow start
but there are 315 days left and the historical base rate is ~15/year (USGS seismic records since 1900)
market expects ~12
data says 15 and 17–19 occurred in 8 of 25 years
2021-2023 each saw exactly 19
'17-19' is at 11¢. model says 25¢ - that's +191% EV
'20+' is at 4¢. model says 9¢ - more than 2x underpriced
'11-13' is at 32¢. model says 24¢ - most overpriced
the market is panicking on recency bias
science doesn't care about the first 50 days
climate markets are full of edges like this if you run the numbers.
Your Edge for Climate Markets on Polymarket (Part 1)
I've been digging through categories nobody talks about to find an edge.
First niche I looked at: Natural Disasters
I've found a market - How many M6.5+ earthquakes will happen in a 7-day window?
Globally there are ~40 M6.5+ earthquakes per year. That's ~0.11 per day.
Not as many as you'd think.
To calculate the probability of each outcome on Polymarket, we use the Poisson distribution - the standard model for counting rare, independent events over time.
λ (expected events) = 0.11 per day × 7 days = 0.77
Then P(k) = (λᵏ · e⁻λ) / k!
Using that formula:
0 earthquakes: ~46%
Exactly 1: ~36%
2 or more: ~18%
Now here's the part that surprised me.
I compared my numbers to the actual Polymarket odds - and they're almost perfectly priced. Within 2-3% on every outcome.
That's probably not a coincidence. Market makers are likely running these same models and constantly adjusting the odds. You're not going to sit there with a calculator and outsmart an MM bot doing this 24/7.
But "almost perfectly priced" isn't "perfectly priced."
I'm going to dig deeper - tracking historical windows against actual outcomes to see if there's a consistent gap between model odds and market pricing.
If there is, that's a real edge. More on that soon.
So is there any edge right now? Yes.
But it's not math. It's speed.
These markets don't update instantly when an earthquake hits.
USGS detects a quake -> data gets published -> traders react -> price moves.
That gap between detection and market reaction is where the money is.
If you know about an M6.5 before most Polymarket traders do, you can buy before the odds shift.
The edge isn't in calculating probability better. It's in having information faster.
Problem is nobody is sitting there refreshing USGS all day.
So I built a bot that monitors global earthquake activity in real-time and sends alerts the moment a qualifying quake is detected.
You get the alert, you place the bet, you're early.
The math won't give you an edge on Polymarket climate markets. Being first will.
I'm going to test this theory myself and share the results once I have enough data.
Rossium retweeted
On-chain analysis is one of the most powerful skills in crypto.
2020 → we knew NOTHING about Uniswap. Learned sniping by copying another sniper.
2022 → saw a former MEV competitor farming Blur. We farmed it too.
2025 → spotted wallets doing arb at HyperEVM launch. We tried it ourselves.
On-chain analysis changed our lives.
You need to TRACK and UNDERSTAND what is going on.

















