Catnib (@Catnib3) profile picture
Joined October 2020
Photo by Catnib (@Catnib3)Photo by Catnib (@Catnib3)Photo by Catnib (@Catnib3)Photo by Catnib (@Catnib3)Photo by Catnib (@Catnib3)Photo by Catnib (@Catnib3)Photo by Catnib (@Catnib3)
Catnib retweeted
I enjoyed chatting with @Sokio8D about the unconventional path that pulled me into crypto trading over 10 years ago, despite an initial total lack of interest in either crypto or trading, the adverse information environment, and various lessons learned. Thanks for having me on!
"We had a balance sheet with hundreds of millions of dollars, and Hyperliquid was just kicking our ass." @yenwod_ led research at dYdX while Hyperliquid took its market. He joins the Insilico Terminal Podcast to talk about why, and what he's building differently at @pascaldottrade. We covered his garage-band HFT start in 2015, pricing the odds an exchange steals your money, cross-border arb out of Mexico City, missing MEV, a third of the book on FTX, and why he thinks prediction markets need scalars, cleaner settlement and better ways to pay for liquidity. 00:00 From a garage HFT project to crypto 04:14 From EMA crossovers to tick-by-tick HFT 10:22 Pricing the odds an exchange steals it all 14:31 From Mexico City arb to a trading firm 27:00 15 exchanges, missed MEV, then FTX 37:55 What made money and what wasted weeks 44:04 Watching Hyperliquid beat dYdX 49:36 Prediction markets and building Pascal 01:00:36 How Pascal works: scalars on Solana 01:08:25 Settlement games, fees and liquidity
2
4
54
2,412
Decentralization is a major reason we built Pascal the way we did. Pascal founder @ivo_crnkovic on the magic of DeFi and our core goal of open access 👇️
1
2
21
823
Your weekly market watchlist on Pascal: 📊 October CPI Tuesday, PPI Friday 🌍 Iran peace talks and ceasefire deadlines 🎮 LoL Worlds finals 🤖 Anthropic IPO developments 🇷🇺 Trump-Putin meeting plans
2
2
21
1,063
Catnib retweeted
Inspired by recent discussion about whether it's possible to learn trading outside an elite firm. If you're the kind of person that reads this project and thinks it sounds fun, imo you'll be able to learn employable trading skills just fine on your own.
51
106
33
1,859
260,477
Catnib retweeted
> Every session is a single append only log, and everything else (the transcript, what the model actually sees, token usage, extension state) is just a fold over that log. @nayshins is a real one for this
Ok can finally talk about what I’ve been working on at Reflection. Introducing Mirror, our own open source coding agent optimized for Beam. I started the mirror project when I first joined Reflection, and it has gone through many iterations since then. Install it today with: uv tool install reflection-mirror github.com/reflection-oss/mi…
3
2
23
2,075
Yesterday marked the end of week 2 of our $105,000 midterms rewards program. Congratulations to @outcomenoble @holy_moses7 and @TheWolfOfPoly, and good luck on the next week. As a reminder, trading rewards are distributed each week through midterms (Nov 4th).
4
2
30
1,059
PSA: In addition to API docs, we also have a public Python repository you can point your agent at. In our testing, even weaker models like Haiku were able to onboard, set up automated trading, and run the example market making code. github.com/Pascal-Research-I…
3
23
1,150
Catnib retweeted
I got more data from PredictIt, Polymarket, and Kalshi to check prior years and other election types and did not see it anywhere else. The over- or under-performance in general elections is almost completely driven by party performance.
1
1
7
507
Catnib retweeted
This is interesting! I saw the same for 2026 using prices 7 days before voting, but not for prior years. xitter.cf/gpstraus/status/209017…
I noticed early in the 2026 primary cycle that prediction markets seemed to underprice female candidates in Democratic races. I thought, surely not, then traders would just realize that and correct for it? But now I have 821 candidates across 186 mixed-sex Congressional elections. I compare candidates' election results to their contract price the day before the election and find that there's miscalibration. Interestingly this only shows up in Democratic primaries. It's very reminiscent of the "Richards Effect" in political polling which @johnmsides, Christopher Stout, Reuben Kline, and others wrote about in 2011. The universe of data is small. Prediction markets are new and this is data from one cycle; we'll see if it persists. Why might this be? Is it about the demographic that trades elections—largely young men—or is there a better explanation?
1
2
1
20
2,323
The CLARITY Act failed to reach the required 60 votes on today's motion to proceed. The market on passing in 2026 dropped overnight from a high of 34% to 15% just before the vote.
2
4
36
1,338
Pascal co-founder @yenwod_ on how systematic and discretionary traders can work together Pascal is for all traders
Extinction. Is that where human prediction market traders are headed? "Algorithms are going to take edge away. There's no question." But @yenwod_ thinks it's not zero-sum, and that humans aren't going anywhere.
2
6
51
27,613
Last week, markets reacted to the Missouri Supreme Court blocking the use of a new congressional map without a statewide vote: 1. Redistricting odds dropped from 85% to 16% 2. MO-5 flipped from 16% to 85% Democrat on the news Missouri is one of four states that passed redistricting laws to boost Republicans, without new census data.
BREAKING: Missouri Supreme Court just blocked the new 7R-1D congressional map. Voters will use the old 6R-2D lines in November.
2
2
1
20
1,855
PSA: Pascal has more Clarity Act strikes than any other venue We also notice that one large December position was closed this morning
1
3
21
1,058
A trader just bought 50k contracts on Fed rate hike by Dec 9th Hike odds have risen steadily following Fed chair Kevin Warsh's inflation comments on August 28th at Jackson Hole Warsh also opened his speech by discussing "hiking" in nature, prompting double entendre suspicions
2
3
37
1,720
Catnib retweeted
If you are trading, it's probably because you have edge If you have edge, you probably care about your costs!
If you look at the order book data... Pascal is the most liquid exchange to trade everything from Fed Funds to the Texas Senate Race
4
3
67
3,939
If you look at the order book data... Pascal is the most liquid exchange to trade everything from Fed Funds to the Texas Senate Race
15
4
1
62
6,331
Catnib retweeted
Replying to @atteonthephone
Hello! @pascaldottrade is transparent and onchain. Fee discounts are published in our docs. You can get a 5% discount using someone's referral code (and our base fee is already lower than Poly's). Happy to help you onboard if you run into any snags!
2
3
1
42
9,731
Catnib retweeted
We are building @pascaldottrade to reify our vision of a high quality venue that treats traders with respect Back when I was running a small crypto HFT, various exchanges (unfairly) provided me with the following: - Fee discounts (and maker rebates) beyond the best published tier - Rate limits past the documented max - Ability to make uncollateralized spot trades and settle t + 1 - Info about undocumented API endpoints - IP whitelisting to skip Cloudflare - AWS VPC peering for better latency I say "unfairly" because these were deals that happened at an exchange ops or bd person's discretion and were not documented publicly, not because there's anything wrong with giving different fees or rate limits according to objective criteria. For example, rate limits are a common resource, shared by all market participants, because the exchange only has so much total throughput available. Therefore it's not unreasonable for the exchange to allocate extra rate limits to big top of book makers -- as opposed to, say, stink bidders -- so they can quote more symbols, because this is beneficial to everyone on the exchange. BUT there is no reason to add friction or secrecy to this process. I think the recipe can be very simple: 1. Publish documentation describing any capabilities beyond the baseline that certain participants can access, eg better rate limits or fees. 2. Publish criteria for inclusion, eg do $_ trailing 30-day volume to qualify for [fee discount | better rate limits | VPC peering | skipping cloudflare]. 3. Make all market data public, including fees paid. If you're onchain (like Pascal), make every account's capabilities public as well. 4. Automate the application of (2) as much as possible! For example, it's not hard to automatically allocate rate limits as traded volume changes, but for some reason it's far less common than eg automatically updating fees. More speculatively, there's no technical reason IP whitelisting to skip Cloudflare or VPC peering cannot be done permissionlessly via the exchange API as well, without first wining and dining the exchange team (there is however some risk of abuse, eg DDOSing as soon as whitelisted, that might require a touchpoint with the exchange team). --- One thing I (personally) always appreciated about trading crypto was that *anybody* who felt like trying their hand could show up and compete, with minimal fixed costs and without needing to be particularly well-connected. Unpublished side deals always rubbed me the wrong way because they erode this virtue of low barrier to entry.
Replying to @atteonthephone
Hello! @pascaldottrade is transparent and onchain. Fee discounts are published in our docs. You can get a 5% discount using someone's referral code (and our base fee is already lower than Poly's). Happy to help you onboard if you run into any snags!
9
6
1
111
7,528
Catnib retweeted
Exchange tech thread: how we run multiple matching engines and upgrade without downtime
18
19
4
236
21,335
Today Pascal has 2282 listed markets on: - Midterms: 1078 (House, Senate, governors) - Sports: 813 (MLB, soccer, tennis, F1, NFL, NBA, esports) - Finance: 134 (rates, IPOs, oil, macro) - Geopolitics: 34 (Iran, Hormuz, Bab el-Mandeb, Lebanon, Taiwan)
7
5
2
60
7,924