artemiso 

There's nothing fake about edge nor backtesting. The simplest form of "edge" manifests when you're posting a passive order. And that's found in the premium over fair value that you're willing to accept contraliquidity at. This can be any arbitrary value that you decide is compensation for inventory, pricing or pricing model risk for carrying the position. If my model forecasts that X is worth $x, the simplest way to provide passive liquidity is to ask for a symmetric edge of +/- some small delta, and bid $x-delta and offer $x+delta. You can only collect this premium when you compress the spread, but most of the time your edge will be wider than the BBO and out of touch. So whatever you use to develop a tighter delta — be it speed, informational advantage, modeling complexity — can be seen as your "edge". (The market is extremely efficient. Even with low latency and sophisticated models, most of the time I'd be quoting outside the BBO — so depending on your sign convention, you could say my edge was very small or even negative.) The reason floor traders don't really exist these days is because computers are so much better at finding fair value and updating their delta. You only need to look at the top 3~ firms that have replaced them: Citadel, Virtu etc. all use simulation and backtesting extensively.

A great post. Even in the pit either the existing bid or offer was better than the edge I was interested in, so I when initiating a trade I was always only on one side of the market. My rule of thumb as a scalper was to initiate with "my" edge and get out in the middle of "my" bid/ask spread, except when taking a money management loss, where I'd give up the edge.

However as market makers we couldn't really just do what we wanted. If there were big bids and offers in the pit you pretty much had to match them or brokers would stop trading with you.

jstnbrg 

I do not trade any more. I've tried a couple times but I need an edge and I've decided I don't have the time or the financial urgency to devote the thousands of hours necessary to develop one. Even with the giant advantage of standing at the center of the trading universe (the floor) it took me three years and a fair amount of good luck before I became consistent enough to make a steady living. I don't need the money; I had a low eight figure career, and what I'm proudest of is that I managed to keep most of my money. Frankly, at age 65 there are things I'd rather do than stare at a screen 10 hours a day. I am still interested in the business. What brought me back to this forum was curiosity about whether the people here are talking about AI and what they're saying. The last time I spent much time here, 5-6 years ago, there was not much discussion of it. But now I read about ChatGPT writing decent code and there are YouTube videos about people using it to trade, and I was wondering whether that comes up here. I gather from your recent posts that you are retired from the HFT universe. I'm as curious about that as lots of others are about floor trader mentality. While I know a couple of the original investors at Getco, they don't talk about it and I don't ask. But I am dying to know how it works, in a general sort of way. I read "Flash Boys" by Michael Lewis but I don't think he ever got to see the real picture; mostly he wrote about arbitrage in a fragmented equities market, and there appeared to be a whole lot more than that going on when I was still trading the yield curve on the screen. For example, I spent a couple years trading out of Advantage Futures satellite facility in Downers Grove, IL around 2005-6 with a lot of other traders and I once heard a guy bragging about how the group he was with had technology that could send out 17,000 messages/second. What would be the advantage of that? Thanks in advance for whatever you're willing to share.

Thanks for the followup.

I think your query was directed at @artesimo - so I let him address it

phantomtrader 

Interesting experience. I think floor traders had the advantage of inside trading. It was a hand signal, a nose twitch, knowledge shared behind the scenes. Don't you think that this really accounts for the "edge"? Since electronic trading came into being, all that went away. The market doesn't give hand signals or nose twitches. But it does give a lot of technical information. To me, order flow is the most important tool for a scalper, which is what I do. You mentioned that you used order flow on the floor. I have no idea what that would have looked like in those days. What did order flow look like in those days? I remember seeing pictures of big boards flashing numbers. How were you able to read the order flow and use it to your advantage? Order flow tools today improve all the time. I couldn't trade without them. It's always good to hear from someone who's been there, done that. Thanks for posting.

We certainly had the advantages you describe. It's why exchange memberships used to cost so much. Firstly, individual brokers tended to represent only one or two large firms, so you got a sense of who was doing what. In my pit there was one guy who's main customer was a big cash arb (basis trader) so I always knew what the basis traders were trying to do, i.e. where their futures bids/offers were relative to the cash. Stupid that they only had one broker, really. In the Treasuries unlike the grains, we could see the whole cash market: bids, offers, size, trades, in real time on Cantor Fitz/Garban screens. Also, you could get a sense of the net position of the guys in the pit by how they sounded; if they were really short and a big offer came into the pit, the volume would go up a lot, with a definite hint of begging My first mentor was a lot more psychologically astute than me and as the market reached new highs or lows for the day/week/etc, he'd watch the brokers to see who was more alert than normal: he was watching for signs they held stops.

The way I used order flow was that I chose to make my living "leaning on" the cash/futures arbs, and I got pretty good at tape reading the cash markets. When I was making my most money the cash was a lot deeper market than Five Year Note futures (that changed later) and by watching how bids and offers in the cash market responded to overall changes in price I could figure out if someone had an ax to grind; for example if a pattern of upticks in futures drew in more offers in the cash, it was a sign that there was a big seller out there and it was relatively safe to have a big short position (because I could "lean" on the offers they had not yet put into the pit).

kevinkdog 

Exactly. In other words, they had an EDGE, 3 edges really. They did not need to backtest, because new floor traders usually had friends or relatives who were floor traders, and they knew those people made money. That was their "proof of concept" (aka backtesting). When markets went electronic, most floor traders tried to transition to "upstairs" but most could not. Advantages a) b) and c) all went away with move upstairs. I bet over the years I have been approached by 50+ floor traders trying to figure out how to reclaim their lost glory...

You might be describing me, except for the last line of your post. Floor trading and modern electronic trading involve completely different skill sets. While I think I'm far from being stupid, I don't have a math/engineering/computer science background, and I think that's the skill set one needs to be a successful electronic trader. Floor trading was primarily a mix of social and self-management skills with maybe a little pattern recognition thrown in. I don't mean to minimize what it took to be a successful floor trader. It's HARD to risk a significant part of your net worth every day.

Actually, I did successfully make the move upstairs initially, by changing what I was trading, (from scalping Five Year Note futures to scalping yield curve spreads, i.e. ZF/ZT, ZT/ZB, ratioed). I was invited to move off the floor by a friend who, along with a former employee of mine and the biggest former broker in the 30 Yr bond pit, ran a prop group of yield curve traders. I wasn't part of the prop group, but I did use and pay for space in their facility, data feeds, etc. and was available to help mentor their employees. I went from being a large (300-500 contract) outright scalper to being the smallest possible FITE (5:3) and NOB (3:2 or 2:1) trader until I could learn to make money consistently. I started making money pretty quickly, and over about 2 years I was able to grow my size trading the NOB to 200:100. But when the CBOT changed its data feed from EUREX, which updated about every 1/2 second, to LIFFE Connect, which was continuous, it suddenly became impossible to execute my spreads. I'd get "legged up" (miss a leg) and either have to scramble to get out of the leg I had executed or get a bad execution on the second leg and have to work just to scratch or take a small loss on the spread I'd just badly executed. I guess the HFT guys needed the continuous data feed, and once they turned their machines on it was "Game Over" for me.

kevinkdog 

Thanks for the followup. I think your query was directed at @artesimo - so I let him address it

Yeah, I went back and looked at the thread again and saw you were responding a post he(?) had made. Sorry. Hopefully @artesimo will see this and respond.

It's @artemiso. You both spelt it wrong!

jstnbrg 

We certainly had the advantages you describe. It's why exchange memberships used to cost so much. Firstly, individual brokers tended to represent only one or two large firms, so you got a sense of who was doing what. In my pit there was one guy who's main customer was a big cash arb (basis trader) so I always knew what the basis traders were trying to do, i.e. where their futures bids/offers were relative to the cash. Stupid that they only had one broker, really. In the Treasuries unlike the grains, we could see the whole cash market: bids, offers, size, trades, in real time on Cantor Fitz/Garban screens. Also, you could get a sense of the net position of the guys in the pit by how they sounded; if they were really short and a big offer came into the pit, the volume would go up a lot, with a definite hint of begging My first mentor was a lot more psychologically astute than me and as the market reached new highs or lows for the day/week/etc, he'd watch the brokers to see who was more alert than normal: he was watching for signs they held stops. The way I used order flow was that I chose to make my living "leaning on" the cash/futures arbs, and I got pretty good at tape reading the cash markets. When I was making my most money the cash was a lot deeper market than Five Year Note futures (that changed later) and by watching how bids and offers in the cash market responded to overall changes in price I could figure out if someone had an ax to grind; for example if a pattern of upticks in futures drew in more offers in the cash, it was a sign that there was a big seller out there and it was relatively safe to have a big short position (because I could "lean" on the offers they had not yet put into the pit).

Thanks so much for the detailed reply. One question: were floor traders able to spoof the way they do today? One of the most annoying aspects of order flow, particularly using Jigsaw, are the fake numbers - here one minute, gone the next, pulling orders, you know the drill. Just wondering if stuff like that went on with floor traders.

Thanks again.

goodoboy 

Good Morning jstnbrg, Thank you for the write up and sharing your experiences. A few questions please: 1. From your memory was floor traders scalping or swing trading from your experience? Scalping meaning taking small profits and swing trading meaning holding the trade position for longer minutes or hours. 2. What was the money mindset of the floor traders? Was it make a living to pay bills or get rich as fast I can to pay off debt, then working about making living? Thank you,

1) Every floor trader I knew was a scalper, many of whom were scalping spreads. But sometimes those "scalps" would take hours to get out of, especially for the spreaders but also when things got slow. The business model for a floor trader is "high volume, low margin, high fixed cost, low per/contract cost".

2) Mindsets varied a lot. Some guys (I'd say 95% or more of floor traders were men, it was a pretty hostile environment for women) would get to where they could make a very comfortable living trading a certain size and then would just keep trading that size, others like myself just kept getting bigger. I started in the Five Year pit as a 5 lot trader because that was the smallest size where a broker would even look at you, and then traded 1 contract/$2000 in my account. Eventually I got to the point where financially I could carry 1000 contract positions, though it wasn't easy to get that many on and still have a trade you could exit quickly. But for a couple years I routinely had 500 contract outright positions. The advantage to trading size wasn't just that you could make a lot of money; it was that brokers wanted to trade with you because it was a lot easier for them to sell 500 contracts to one trader than to divide that up among 10-20 traders and risk losing their count. Brokers were financially on the hook for any mistake they made. They made ~80 cents/contract but if they lost a single tick on a mistake they'd be out ~$15.62 per contract. One time I had an outtrade with a broker in a fast market where it moved 35 ticks before we figured out the mistake three minutes later.

Some traders never got large. For years I stood between two guys who wouldn't take more than 10 contracts. Another guy I knew said that the $70K/year he made was "pretty good for a guy from Rockford".

Finally, with one exception the big gambler types mostly blew out. The exception was Harris Brumfield, who went on to found Trading Technologies. That guy seemed constantly to be at war both with the market and with most everyone else trading. But he was a great trader. Most floor traders were actually pretty skittish.

phantomtrader 

Thanks so much for the detailed reply. One question: were floor traders able to spoof the way they do today? One of the most annoying aspects of order flow, particularly using Jigsaw, are the fake numbers - here one minute, gone the next, pulling orders, you know the drill. Just wondering if stuff like that went on with floor traders. Thanks again.

Yes and no. If your hands were up and a broker said "Sold" you couldn't back out, both by the rules and by the social conventions. I showed big offers all the time when I actually wanted to buy, but I'd NEVER EVER back down from an offer I'd just made. It was professional death to do so. Also if I was making that kind of an offer I'd have been getting a good edge for the trade, so most of the time it would be OK anyway.

I think bluffing is part of the game. Ignore the spoofs, just do what you want to do. Also, sometimes those "spoofs" may be many spreaders all leaning on something else. I remember many years ago watching a 5 lot offer in a back month generate 800 lots of selling in the front month, presumable as 160 different traders' Autospreaders all tried to execute the same spread.

OK, gotta go. Heading to Big Bend National Park tomorrow and there's laundry to do.