Conversation

2) FTX wasn't really founded on the notion that it was going to make better tradeoffs than other exchanges. It came from the belief that there were big wins to be had by just doing things better. That to the extent there was a tradeoff, the sacrifice was hard work.
2
251
3) And, in a lot of ways, I think this was true. Margin engines in 2018 were really bad--losing a million dollars per day. Regulatory approaches were naïve and risky. Customer support was spotty at best. We could do better.
5
161
4) And I'd like to think that we *did* do better--and that in doing so, we moved the industry forward. 'Clawbacks' are no longer an everyday occurrence--they're rare, are rightfully hated. The industry has started engaging in productive conversations with regulators.
2
152
5) In other words, there was slack in the system. We could make significant improvements in some areas of the product without sacrificing other areas; all it took was a lot of work and thought and iteration and feedback.
1
138
Replying to
7) I'm happy with the decisions we've made--and I'd make them again. The single most important thing for us, and for the industry, was to build constructive, respectful relationships with regulators, and that's been our single biggest focus.
4
177
8) And, as such, FTX is now licensed in more countries than any other exchange in crypto--dozens more. And we'll keep pushing forward, working to build out safe, robust regulatory frameworks, and to comply with them. Because that's the most important thing for the industry.
5
193
9) But it has come at the cost of a *lot* of time--particularly dev time. Behind the scenes, we've spent thousands of hours this year building out more and more robust compliance, reporting, KYC, and localization systems. Meanwhile, our market share grew this year--a lot!
Image
9
226
10) But as our volume grew, we used up the slack in our system--in this case, the slack in our matching engine. Over the last few months, we've been hitting system max processing capacity more and more. And probably relatedly, our market share stagnated a bit.
Image
7
118
11) And, while it was an imminently solvable problem, it was one that would take a lot of careful work--work we put off, possibly longer than we should have, to care for other priorities.
2
102
12) Anyway, over the last few months we've put a ton of time into reducing latency and increasing throughput. And, this week, we finally started rolling out the upgrades. Thanks, to everyone who pushed us to get better; and sorry, that this didn't come sooner.
4
181
13) So, users should now see: a) greatly reduced fill/position/balance reporting latency b) ~10ms lower order placement latency c) significantly increased order matching capacity There are more upgrades coming, but the first batch are live.
7
237
14) Along with this, we're also making some changes to ratelimits. Going live Friday (Jan 21st) at 1pm EST: a) VIP/MM ratelimits will be increased by roughly 10% b) VIP/MM ratelimits will only apply to orders above a certain size threshold (~$1k-$5k), depending on the market
7
125
15) This should substantially increase throughput and decrease latency, especially when markets are the busiest. For website/app/other tiers, ratelimits will either be unaffected or increase. We'll put the full updates here soon: help.ftx.com/hc/en-us/artic.
9
93
16) Anyway, there's more coming--and hopefully we'll be able to take ratelimits up further soon. For now, we might have a *little* bit of slack in the system. I expect it to be used up in short order, though. So: back to work.
27
176