Why strategies make money

[garachen]

When I am thinking about a new trading strategy I've found it helpful to at first focus on the 'why' this strategy should make money before getting to the 'how' the strategy makes money. And if someone is explaining their strategy to me and can't answer the question of why the strategy is supposed to make money I immediately become pretty skeptical.

Trading is more than pattern seeking. Looking at patterns is helpful to identify how to make money but first you should identify why your strategy deserves to get paid - what service are you providing the market that justifies your income - and then look at patterns to then identify when this service you provide has greater or lesser value.

Here's a few of the 'why's' I can think of: (futures)

1) Providing "smart" liquidity. Stepping aside during toxic flow but taking the other side of uninformed traders. A very good strategy here makes a little less than 1/4 increment per round trip.

2) Simple information dissemination. This is the speed game. New information comes out. Economic release, crop report, pure arb, etc and prices need to move to adjust to the new reality. You serve the market by adjusting the price.

3) News dissemination. Complex news comes in many formats. Not everything is coded up to be machine readable. Standard news needs people to interpret and act. Usually you have about 2 seconds for standard products and maybe up to 10 seconds for less obvious products. Really esoteric news/product combinations can disseminate over a period of hours.

4) High correlation microstructure relationships. This is actually a good place to be in. You need to be fast but not overwhelmingly fast and there are tons of these relationships. Think of things like the energy or ag complex. Lots of relationships and short term correlations. Looks for iceburgs or stops or other short term disparities.

5) Medium term 'macro' relationships. Sometimes these get out of whack and someone needs to bring them back inline. Need to have a good understanding of the actual products to not get killed and a pretty big balance sheet.

6) Calming the panic. People and markets tend to over-react to news or to extreme price movements. They may be forced to exit a position either by their own risk tolerance or by policy at the end of the trading session. These episodes of panic that are purely price driven tend to be short lived and you can get paid to 'calm the storm'.

I think that no matter the specifics of your strategy you should always be able to answer the question of what you are providing that market that justifies you getting paid for your efforts. If not one of these reasons then something else that provides value.

Out of the 6 'whys' I listed, I believe two are suitable for retail traders (people at a fee/speed disadvantage). After capitalizing on those two you should have enough $$$ to pursue the other 'whys' which generate more stable income.

Which two to start with?

 

[FCXoptions]

Out of the 6 'whys' I listed, I believe two are suitable for retail traders (people at a fee/speed disadvantage). After capitalizing on those two you should have enough $$$ to pursue the other 'whys' which generate more stable income.

Which two to start with?

First off, thanks for posting this. The other day I was working on something, and could see it was working a large portion of the time, but still was taking some larger losses at times. Your post reminded me to think about why it is working more than the fact it is working which helped with the process quite a bit.

I would say #3 and #6, simply from the fact the rest probably require speed, a big account for it to work well, or significantly reduced commission. Although #3 I guess you still need to be pretty fast from a human standpoint to quickly figure out what to do based on the news.

Hope you are doing well and had a good Thanksgiving!

 

[jtrader33]

I would say #3 and #6, simply from the fact the rest probably require speed, a big account for it to work well, or significantly reduced commission. Although #3 I guess you still need to be pretty fast from a human standpoint to quickly figure out what to do based on the news.

Agree with your picks, although as an equities guy I'm surprised there are enough opportunities in either category in futures to make them interesting as a foundation for a trading business.

 

[londonkid]

Another reason strategies work: your counterparty isn't or can't be 100percent profit oriented on the same risk factors you are.

Great post. Would you care to share an example. Of course i understand if you don't want to.

 

[garachen]

First off, thanks for posting this. The other day I was working on something, and could see it was working a large portion of the time, but still was taking some larger losses at times. Your post reminded me to think about why it is working more than the fact it is working which helped with the process quite a bit.

I would say #3 and #6, simply from the fact the rest probably require speed, a big account for it to work well, or significantly reduced commission. Although #3 I guess you still need to be pretty fast from a human standpoint to quickly figure out what to do based on the news.

Hope you are doing well and had a good Thanksgiving!

Much of my Thanksgiving is held in a language I don't understand.

Yeah. 3 and 6. You need to start with something that generates a high enough $/trade to overcome your commission disadvantage. Then use your success there to buy down your fees elsewhere.

So, at first you should be in a place where algos have a disadvantage because it's a suboptimal environment for them.

 

[Gambit]

Much of my Thanksgiving is held in a language I don't understand.

Yeah. 3 and 6. You need to start with something that generates a high enough $/trade to overcome your commission disadvantage. Then use your success there to buy down your fees elsewhere.

So, at first you should be in a place where algos have a disadvantage because it's a suboptimal environment for them.

What do you think of liquidity provision on a longer time frame? Putting limit orders much deeper in the book and combining that with longer term mean reversion?

 

[applejuice]

Very much enjoy garachens posts but re: "path towards profitability"
* "Try to gain intuition to determine if there is selling pressure or buying pressure."
* "not super news sensitive"
* "But I would be curious if anyone were to follow my method to fruition."
Difficult to reconcile this "path towards profitability" with #3 and #6. That path feels more like developing intuition for the ebb and flow rather than grabbing ticks on breaking news and spikes in VIX and gold.

I am in awe of the people who are able to generate $5k/day from price ladders with <1 year experience. I used to literally sit 12 inches from my screen and block out everything else in the world except for the price ladder of one or two major Eurex products but must have been missing the magic ingredient.

IMO if you want to trade news to build an account you need a bloomberg or reuters terminal to have a shot. I've sampled from the major Squawk boxes and the news which moves markets violently enough, was in my experience infrequent (<<1/month). With a BBG terminal probably would have gained the 8 second head-start on the Squawk boxes and made ~3 semi decent news trades/month.

If I had my time again, I would have clicked buy/sell as soon as I saw the markets starting to run + the "click" of the squawkers microphone. Instead I waited around to try and judge the news and miss too much of the smaller moves.

 

[garachen]

Very much enjoy garachens posts but re: "path towards profitability"
* "Try to gain intuition to determine if there is selling pressure or buying pressure."
* "not super news sensitive"
* "But I would be curious if anyone were to follow my method to fruition."
Difficult to reconcile this "path towards profitability" with #3 and #6. That path feels more like developing intuition for the ebb and flow rather than grabbing ticks on breaking news and spikes in VIX and gold.

I am in awe of the people who are able to generate $5k/day from price ladders with <1 year experience. I used to literally sit 12 inches from my screen and block out everything else in the world except for the price ladder of one or two major Eurex products but must have been missing the magic ingredient.

IMO if you want to trade news to build an account you need a bloomberg or reuters terminal to have a shot. I've sampled from the major Squawk boxes and the news which moves markets violently enough, was in my experience infrequent (<<1/month). With a BBG terminal probably would have gained the 8 second head-start on the Squawk boxes and made ~3 semi decent news trades/month.

If I had my time again, I would have clicked buy/sell as soon as I saw the markets starting to run + the "click" of the squawkers microphone. Instead I waited around to try and judge the news and miss too much of the smaller moves.

I think market conditions have changed since my original post on that other thread. Even a few years ago it was relatively easier to find products you could manual trade and grind out money daily.

Nowadays, I think there's better opportunity in waiting, trading less, and only stepping in when the odds are greatly in your favor. That means very few trades per week and in moderate to large size.

With low fees, the daily grind still works but with retail fees its harder to find the appropriate contracts and time of day than it used to be.

Also, by trading on complex news I don't mean anything that is regularly scheduled. I mean random news whose primary source might not be English.

 

[Rum Runner]

I used to run a completely automated #1 in the equities market. It ran all day long in a couple of dozen symbols. For a couple of years it did well, but the returns are no longer there, so I've moved on. I'm now running a mostly automated (some trader input is needed throughout the day) that is a combination of #1 and #6 in futures. It starts making markets only in times of panic or over reaction. I've found this combination to be much more reliable than having it run at all times of the session with 100% automation.

 

[garachen]

What do you think of liquidity provision on a longer time frame? Putting limit orders much deeper in the book and combining that with longer term mean reversion?

It depends on several things.

How thick is the book --> how much effort do you need to put into calculating the value of your position in line.

How are you deciding when to cancel orders

How are you managing the exit. And are you treating profitable or loss exits differently

What effect are you trying to capture. A typical range mean reversion or a mean reversion due to abrupt overreaction.

Think through these and then attempt to sim to see if $/trade is high enough for your cost structure. Or just run and find out. Maybe expensive to find out by running but at least it's accurate.