I’ve been intrigued by the world of quantitative trading for many years, so when ex-Jane Street trader Ricki Heicklen ran a trading bootcamp at Manifest recently, I jumped at the opportunity! The bootcamp was a two-day intensive program designed to teach the fundamentals of quantitative trading and provide a glimpse into the experience of being a professional trader. Using a play currency, we went through various hands-on exercises to solidify concepts taught in energetically-delivered lectures. I learned a lot and made friends, and had so much fun that I’m considering a career pivot to quant trading. So, this post has two purposes: to give interested readers some background in trading concepts and a sense of my bootcamp experience, and to serve as a proof of work showing potential employers what I learned.

Understanding Markets

To understand how to trade in a market, we first need to understand what a market is and how it functions. A market is a platform where buyers and sellers come together to exchange goods, services, or financial instruments. Market structures vary widely, but the bootcamp primarily focused on the standard structure for US equity markets, and that’s what I’ll be talking about here. At its core, the market’s functionality is driven by the concept of an order book.

An order book is fundamental to financial markets. It lists all open buy orders (bids) and sell orders (asks or offers) for an asset, sorted by price and then by time. This means that Markets operate this way to ensure fairness and liquidity. Sorting by price gives priority to those who provide the best price and gives incoming traders the best available deal automatically, and prioritizing older orders encourages placing orders quickly rather than waiting for others to place orders first or constantly canceling and replacing orders. . For example, an order book for some stock (let’s call it Shmoogle) might look like:

Bid Trader Bid Time Bid Size Bid Price Ask Price Ask Size Ask Time Ask Trader
Joshua 10:13:30 100 50.02 50.04 150 10:14:22 Michael
Max 10:15:45 250 50.02 50.05 200 10:13:05 Stefanie
Andrey 10:14:20 300 50.01 50.06 100 10:12:30 Yoav
50.07 250 10:11:55 William

In this order book, Joshua, Max, and Andrey want to buy Shmoogle, while Michael, Stefanie, Yoav, and William want to sell Shmoogle. Coming into this market, you could sell up to 100 shares to Joshua for $50.02 each, or buy up to 150 shares from Michael for $50.04 each, before moving on to the orders below theirs in the book. Notice how Joshua gets to trade before Max at the same price, because he submitted his order first. You could also place your own bid or offer, and if you were willing to trade at $50.03 (in either direction), you’d get priority when someone new wanted to trade!

The best (highest-priced) bid is always at a lower price than the best (lowest-priced) offer; if this changes, the book is called “crossed” and one or more trades immediately execute. For example, if you tried to place a bid to buy 350 shares for $50.06 each in the book above, instead you would immediately fill Michael and Stefanie’s existing orders by buying Michael’s 150 shares for $50.04 each and Stefanie’s 200 shares for $50.05 each.

I was taught some specific jargon for orders, used for trading exercises in the bootcamp and, I presume, for real trading in the olden days before it all became computerized. For example, continuing to trade Shmoogle:

  • "price bid for size": Place a bid to buy up to size shares of Shmoogle for price or lower each (e.g., “10 bid for 5” means buying up to 5 shares at $10 each).
  • "size offered at price": Place an ask to sell size shares of Shmoogle for price or higher each (e.g., “5 offered at 10” means selling up to 5 shares at $10 each).
  • “I’m out”: Cancel open orders in the Shmoogle market.

To get comfortable with the mechanics of order books, we played a game called Tighten Or Trade. We established an order book, written by hand on a whiteboard, for a contract that would settle to the total number of siblings of people in the room. That is, if it turned out that the people in the room had a total of 17 siblings, the value of the contract would be $17, and anyone who had bought it for $15 would profit $2 per share. With the market established, we took turns either “tightening” the order book by placing a bid or ask better than the current best price, or trading with an existing order from the book by either buying from an existing offer or selling to an existing bid. In fact, after a few rounds of turn-based Tighten Or Trade, we had a short period of live trading on the market where we shouted out orders without taking turns as people gradually revealed their sibling numbers, which better simulated the chaos of a normal market. we resolved the market by counting up our siblings; those who sold at more than the true price or bought for less profited, while the people on the other end of those trades correspondingly lost out. In my case, I thought people were undervaluing the contract so I bought 2 shares for $20 and 1 share for $21. I turned out to be correct, and profited $14 when the contract settled to $25! By interacting with the market in a simplified, turn-based game, we were able to get a feel for order book dynamics in a low-pressure setting with time to think and ask questions.

Adverse Selection

One session of the bootcamp focused on adverse selection. Frankly, if you have the time, you should skip this section and read Ricki’s blog post on the topic instead. Otherwise, here’s my summary:

Adverse selection occurs when your available choices turn out to be worse than they initially appear, because other people selectively take the options you want and leave the ones you don’t. Although especially salient in the competitive zero-sum world of financial trades, it appears in a wide range of other contexts as well. To borrow an example, imagine asking a friend to toss you a Laffy Taffy candy from a bowl at a party. Like most people, you dislike banana flavor, but you figure a 75% chance of getting one of the other three flavors is pretty good, and you don’t want to seem picky. Your friend reaches into the bowl and tosses you a Laffy Taffy; it’s banana. What went wrong? Unfortunately, because other people have selectively been taking their preferred flavors from the bowl, much more than 25% of the remainder is banana flavored. The option you ended up with is precisely the option other people turned down. In trading, adverse selection means that the open orders on the book probably aren’t very good (or someone would have traded with them already) and that among orders you place, the ones that are bad for you are more likely to be traded with than those that would be good for you.

Given the risk of adverse selection, it might seem that people should never accept trades at all. However, there are several factors that can mitigate adverse selection. A non-exhaustive list:

  • Naïve counterparties: If someone doesn’t know what they’re doing, they can take worse options and leave you with better ones (no one else at the party has tried Laffy Taffy before; someone’s uncle Bob has a good feeling about a stock).
  • Unpopulated markets: If there aren’t many people selecting options, or you get to choose before most people, the good ones can stick around (you got to the party early; you have a server farm on the NYSE trading floor).
  • Anti-correlated preferences: If what you want is unusual, the options other people take won’t be the ones you hoped for (you happen to love banana flavor Laffy Taffy; you want to control a company even if it loses you money).
  • Constrained counterparties: Legal or other restrictions can force people to take worse options than they otherwise would, leaving better options for you (everyone else at the party follows a niche religion that prohibits eating non-yellow candy; legal limits on risk force a bank to sell an asset that’s profitable in expectation).

It’s critical to have a story for why adverse selection doesn’t apply in your case before making a trade. That story may look like one of the above factors, and takes one of two forms:

  • “Here’s why my counterparty is wrong about this trade”: In a zero-sum trade, for you to win, your counterparty must lose. If they’re willing to trade, they presumably think they’ll win, so only trade if you have a good reason to think they’re wrong. Note that this can be recursive; if your counterparty is sophisticated, they have a story about why you’re wrong, and your story must explain why their story about you is wrong.

  • “Here’s why this trade is positive sum”: In a trade where both parties benefit, no one needs to be wrong in order to trade. This commonly appears as comparative advantage (e.g., when you buy a T-shirt, you aren’t being misled about the value, the manufacturer can simply make it more cheaply than you could). While positive sum trades are very common in life, they are rarer in financial markets.

If you don’t have either kind of story for why a trade is good, walk away.

Inattentiveness

One implication of adverse selection is that it’s dangerous to make trades inattentively, because your bad trades will be snapped up while the trades that would have been good for you won’t get filled. A few cautions to keep in mind:

  • Never place market orders! A market order says “trade with whatever orders are on the other side of the order book, no matter their price”. If you don’t pay close attention to the order book, a market order may trade at prices vastly different than you hoped, as in a recent example from If you're interested in finance, I highly recommend subscribing to the newsletter. Matt Levine tells some fascinating stories! wherein a five-figure market order ends up on the hook for $74 million. Instead, place a limit order, which won’t execute if the price is worse than you specify. When it executes successfully, it’s the same as a market order, and when it doesn’t execute, you didn’t want it to.
  • When placing limit orders, set short expiration times (ideally immediate-or-cancel) unless you have a very good reason to leave an order open. If your unfilled limit order doesn’t expire, it will be filled precisely in the case where market conditions change such that your trade will lose money.
  • If you have an automated trading system, pause it and cancel your orders when you can’t keep an eye on it to make sure it’s performing within expectations. Remember, if you don’t trade, One of my helpful draft reviewers points out that the value of assets you already have can go down if you don't trade. That's true, but remember adverse selection; even if you think the price will fall, you only get to sell the asset if someone disagrees!

Arbitrage

Arbitrage is a way to make a guaranteed profit by noticing when markets get out of sync in how they value things. To learn and practice this skill, we had a live crossword competition with several markets on the outcomes. There were two teams, Green and Orange, and in addition to the market on which team would win, there were markets on the fastest solve time for each team (GTIME and OTIME). The latter were key, because there were also markets on the sum of the times (SUM) and the difference between the green time and the orange time (DIFF). Bots were created to trade in all the markets to provide liquidity by allowing profitable trades against them, and the stage was set for a lesson on arbitrage.

As the prices of the markets drifted, arbitrage opportunities appeared. For instance, I'm handwaving a bit here, because there isn't actually such a thing as "the price". When financial news reports a single number, it's often the midpoint between the best bid and best ask or the price at which the most recent trade executed. However, if you want to actually trade, you have to fill an order on either side of the book, rather than magically getting the midpoint price. This difference means you get a worse price and is known as "crossing the bid-ask spread". In this arbitrage example, you would have to compare the ask price of SUM with the bid prices of GTIME and OTIME. In this case, a sharp trader could buy a position in SUM and sell positions in both GTIME and OTIME. Then, regardless of the actual times, the trader would make money. Because changes in the prices of the trader’s positions perfectly cancel out, there is no risk, and the profit is locked in. The same logic applies to DIFF, where one could buy DIFF and perfectly hedge (cancel out price movements) by buying OTIME and selling GTIME.

In real-world markets, hordes of traders keeping a watchful (and often automated) eye for arbitrage opportunities keep market prices largely self-consistent. For instance, let’s look at exchange traded funds (ETFs). An ETF functions like a basket of various assets, e.g. tech stocks. In practice, because the providers need to keep costs down and make money, these exchanges can usually only happen in bulk and a small fee is taken in each direction. For instance, a provider may offer 99 shares of ETF for 100 of each underlying asset and 100 of each underlying asset for 101 shares of ETF. Prices of ETFs tend to closely track the prices of the underlying assets, even though there is no rule enforcing this property. This financial magic works by arbitrage; if the price of the ETF got too high/low, arbitrageurs would make money by buying/selling the underlying assets and selling/buying the ETF until their trading brought the prices back in sync. Similar logic keeps prices of assets consistent across different exchanges, even in different countries, to the extent that assets can be moved between them to correct differences.

Electronic Trading Competition

In the final session of the bootcamp, we participated in an electronic trading competition. Several markets were set up as simplified models of stocks, bonds, and other assets. An API was provided for fetching information about the markets and placing trades on their order books, and we were tasked with creating bots that would trade in the markets. In order to provide liquidity to the markets, and avoid the adverse selection concern of trading with only the other sophisticated bootcamp participants, several bots using naïve strategies were created as part of the competition structure.

With only 2 hours to build our bots, it was critical to quickly figure out which markets were worth focusing effort on, just as in real-life trading. Unfortunately, In retrospect, although there are often great advantages to teamwork, I think the other programmer and I both could have done better by working separately. "What one programmer can do in one month, two programmers can do in two months." -Fred Brooks and we weren’t able to build much of a coherent trading strategy at all. Compounded with technical difficulties in the system running the markets and API, this meant our team eked out only a small profit with mostly-manual trades against the naïve bots. In fact, the top profit in the competition went to a highly-ranked Manifold trader who declined to build a bot at all, simply trading manually while the rest of us tried to get our bots working. So, while the electronic trading competition didn’t give me quite the experience I’d hoped for in building a trading bot, I learned an unexpected valuable lesson: having anything working at all outperforms delayed attempts at more sophisticated strategies. Start simple! You can always adapt by adding complexity later if conditions demand it.

Conclusion and Ongoing Learning

To summarize, here are the key lessons I learned:

  • Beware Adverse Selection: Have a story for why your trade makes sense in a competitive environment.
  • Stay Vigilant: Inattentive trading is dangerous, not trading is always safe.
  • Focus Quickly: Learn to make snap judgements of where further effort will be most valuable.
  • Spot Arbitrage: Inconsistencies between prices mean free money if you find them.
  • Start Simple: A working strategy beats a perfect plan that’s not ready.

Overall, the Manifest Trading Bootcamp was a fantastic experience. I learned a lot about trading and had fun doing it! I hope you enjoyed this write-up, but there’s really no substitute for hands-on experience. Many thanks to Ricki for taking the time to run workshops and teach what she knows!

Post-bootcamp, I’m continuing to learn about trading in a few ways:

  • Some of the other bootcamp participants and I started a book club, and as of this writing we’re reading The Laws of Trading by Agustin Lebron. So far, I’m really enjoying the book! If you want to learn more, there’s a good review of the book on Astral Codex Ten.
  • Since Manifest, I’ve been more active on Manifold, honing my instincts for modeling the world and understanding market behavior in a low-stakes environment.
  • As well as trading manually, I’m developing a Manifold bot in order to learn more about creating automated trading systems.

I hope you enjoyed this post, and as always, please feel free to get in touch!