Our third Algo: Aggressive Liquidity Seeker

TL/DR: Proof has recently launched our third algo. Like our flagship Proof algo, the new algo has two components that operate in parallel: a liquidity seeker and a scheduler. Both components are designed to behave more aggressively than the Proof algo, and we expect it to achieve participation rates around 20%.

Signature Behaviors

• Aggressive liquidity seeker with a “target” participation rate of 20%
• Participation can be as low as 10% or as high as 100% if it finds hidden midpoint liquidity (i.e. can trade as a block in-line at any point)
• The tactical logic (routing, reshuffling, venue, order type selection, etc.) is the same as our Proof flagship algo with a more aggressive top-level scheduling
• Medium minimum quantity for the liquidity seeking component, prioritizing speed of execution while still filtering for toxic counterparties

Highlights of Our Model

• Continues to leverage our homegrown dynamic volume prediction model and optimizes the scheduler to meet a target POV with 2 additional features: Quick Start and MinPOV
• Quick Start: Crosses the spread at time of order receipt if marketable
• MinPOV Tracker: works as an override on top of our scheduler, in case the scheduler falls behind a minimum POV, the algo allocates additional shares that are eligible to cross the spread
• Algorithm is configurable to dial aggressiveness level on a per client basis
• Algo protection: In case market data is considered suspicious by our trading engine the algo will fallback to use a more passive approach

Algo Suite Snapshot

Check the comparison below to better understand the differences between our Flagship and our Aggressive Algo

An Interview with Proof’s Newly Launched Aggressive Algo

“Let’s get this press conference started,” the aggressive liquidity seeker says brusquely, seating himself in front of a microphone and adjusting it with his still-wrapped hands. “We’ve all got places to go, right?”

The press is still filing in, a little dazed from the abrupt ending to the boxing match in the middle of the third round. A few are still holding half-drunk beers. One of them shouts out an easy question to buy some time: “Tell us about your strategy in this fight. Did you plan to finish it this quickly?”

“Yeah, it was about what I expected,” the aggressive algo answered. “At the beginning, I always size up the other guy and decide how long the fight should be, assuming that I participate in landing punches at my target rate. I like to call this my POV, or ‘percentage of violence.’ In this case, I decided about three rounds would be enough. If my corner man holds me back with limits, it may take longer. But that didn’t happen this time.”

Another reporter finishes his beer and lobs a question. “Once you know how many rounds you expect to need, how do you think about spreading your energy over those rounds?”

“Well, you always gotta be looking for the knockout opportunity,” he answers. “If I can get in, right through the midpoint of his guard and land a big hit, I’m always going to take that. But if not… well, I’m going to fall in line with the general rhythm of things. As the fight goes on, I’m tracing the curves of his movement and anticipating where I need to be to keep pace. I’ve studied a lot of recent fights, so I know what typical movement patterns are, and I can adjust based on what I’m seeing in the moment. I call this Dynamic Volume Prediction — dynamically predicting the volume of punches as they come, as a joint function of historical knowledge and real-time observations. Of course, it doesn’t work perfectly. If he’s landing lots of punches within my limit zone and I get behind, I’m going to get more aggressive to catch up.”

One reporter leans against the wall, her arms crossed skeptically. “The gym where you train — Proof Training — has previously produced a flagship fighter that they are very proud of. Why do you think you can be better?”

“Look, I’m not here to trash talk my training partners,” the aggressive algo says, and then pauses for effect. “But…”

The reporters chuckle.

“But that classic Proof approach is only best in situations where you want to be patient. Is the flagship algo a great fighter? Sure. Does he minimize the damage he takes? Sure. Does he have more teeth left than me? Absolutely. But sometimes you don’t want careful — you want scrappy. You want somebody who throws a cross as soon as the first bell rings.” [He punches the air to illustrate the point.] “You want somebody who’s going to get in as many punches as they can when they get on the inside. You’re there already right? Why not go for an oversize combo if your opponent isn’t backing away?” [He throws a hook-cross-hook combination dangerously close to the microphone.] “And if you can complete the whole fight with an inside combo, why not go for it?” [In conclusion, he bats the microphone and it falls over, causing all of the reporters to momentarily cover their ears.]

As a nervous sound technician resets the mic, a reporter shouts: “What about your next fight? I heard a rumor that you were lowering your minimum standards to help find new opponents.”

“Again, I don’t want to talk trash,” the aggressive algo began conspiratorially. “But have you seen the minimum qualifications that the flagship algo imposes for a big match? I mean, sure, you want to ensure high quality opponents, but you’re going to fight the guy, not marry him! Don’t get me wrong, I don’t have low standards. But mine are a bit lower than his.”

“Just one last question,” another reporter calls out. “Aggression is fine when you’ve got your wits about you, but if you notice that your reaction time is slowing or something’s wrong with your perception, what do you do?”

The aggressive algo sighs. “It doesn’t happen often, but in that case, I default to more patient tactics.” He grips the microphone stand tightly. The technician watches him anxiously. “I really hate that though,” he says, still squeezing it. But then he lets it go and smiles. “I’m starving,” he says to no one in particular. “Anybody know where I can get some pizza and beer around here?”

“You might want to shower first,” a front row reporter suggests. “And you’ll need to report to CAT — the consolidated audit trail of fight statistics.”

“Nah, I’ll do that later,” he says.

If you liked our post or have more question/feedback about our algo design please feel free to contact us directly at info@prooftrading.com. Thanks.