Gambling v's Trading

[firlandsfarm]

I know there's a 'reading' thread here somewhere but I couldn't find it and I convinced myself maybe it is an interesting topic in itself. It should give newbies food for thought and even us longer in the tooth are never beyond an occasional little reminder. Anyway I received this email recently and thought some may find the content useful.

From: info=[email protected] <info=[email protected]> On Behalf Of The Trading Cafe
Sent: 21 January 2025 20:33
To: xxxxxxxxxxxxxxxx
Subject: Gambling VS Trading

There’s something we noticed in The Trading Academy and we needed to let you know about it.
A student (Krupal) posted the following:
***BEGIN POST***
Hi Team,
I am slowly but steadily progressing towards the modules, but as I am progressing forward, I am feeling more enlightened as what has been mentioned by Zack in the Gambling vs Trading Part 1 all the six characteristics apply to ME! and I have no shame admitting it. It was one of the enlightening moments for me to realise what grave mistake I have been making since I started trading or should I say GAMBLING
As I'm writing this post, I have open trades in my Live account which would be the last trade. I will go through the process of Learning, doing the Reading, Writing the GOALS, Posting in the community, and improving my mindset till I get to the point where instructors tell me I am Ready for TRADING!
I will keep practicing when I get to that module, till I get it right, revise and watch the videos again till it is embedded in me as my second nature.
Apologies for the long post, but if anyone out there is in the same boat as I am please do tag along.
***END POST***
Most of the time, we don’t realize that we’re gambling. As a trader, there’s a fine line between taking a calculated risk for the sake of education and outright placing a bet in the markets.
Inside of The Trading Academy students get access to The Psychology Care Plan. It’s a simple 10-page document that allows you to identify psychological weak points and fix them.
Here’s what it says about Gambling VS Trading.
Ask yourself the following questions if you’re worried that you’re gambling:
1. Are you finding yourself making impulsive decisions or taking high-risk trades?
2. Do you sometimes lack a clear plan or trading strategy and make decisions based on emotions or hearsay rather than rational analysis?
3. Do you disregard stop-loss orders?
4. Have you ever failed to place a stop-loss order or chosen to override it due to a belief that the market will turn in your favor?
5. Have you failed to diversify your investments?
6. Have you ever invested all your capital in a single asset or market without considering the potential risks and benefits of diversification?
7. Are you chasing losses?
8. Have you ever attempted to recover losses by increasing your risk exposure or taking trades outside your trading plan, leading to further losses?
How to avoid gambling as you learn
1. Develop a trading plan: A trading plan helps you to avoid making impulsive decisions based on emotions or gut feelings. It should outline your entry and exit points, risk management strategy, and other important details. Make sure you fill out, use and adjust your trading plan on a regular basis.
2. Use technical and fundamental analysis: Use data-driven analysis to identify potential trades and make informed decisions. Avoid relying on rumors, speculation or insider information. Stick to the education you receive in the academy and do not stray from it.
3. Set risk management parameters: Define your stop-loss and take-profit levels, and determine the maximum percentage of your account you are willing to risk on any given trade. This helps you to control your risk and avoid taking trades with a high potential loss.
4. Avoid overtrading: Resist the temptation to trade excessively, which can lead to impulsive and irrational decision-making. Set daily or weekly trading limits and stick to them.
5. Keep a trading journal: Record your trades, including your rationale for entering and exiting a trade, the outcome, and lessons learned. Analyze your journal regularly to identify patterns and areas for improvement.
6. Focus on the long-term: Avoid the desire to make quick profits and instead focus on building a sustainable trading career. A long-term approach helps you to stay disciplined and avoid taking excessive risks.
7. Manage emotions: Trading can be emotional, and it's important to manage your emotions effectively. Stay calm and rational, avoid trading when you're feeling overly emotional, and take breaks when needed to avoid burnout.
The truth is that most people will read this and nod their heads internally. You know that you need to do these things. The execution of it is a different story.
Ask yourself this question: do you want to become a legacy trader?
If you do, then you have to do everything in this list diligently.

Note: The Trading Cafe is a financial based trading academy but hey ... trading is trading. I am not connected to them in any way and have no idea if their content is good, I just receive the emails and read them.

[jimibt]

not going to address every item on that common sense list, but i can say from personal, ongoing experience that setting and sticking to stop losses is probably the most liberating habit that i've formed over the past 2-3 years. I actually in fact take comfort when my stops are hit as they prove that you were prepared for the market trading against you (rather than gripping on and hoping that things will turn in your favour).

Back testing on real tick data has been the strongest confirmation to sticking to these plans for me. I will say tho, this is related to forex/commodities/indices, rather than BF, but the same principles apply.

nice list tho...

[firlandsfarm]

Back testing on real tick data has been the strongest confirmation to sticking to these plans for me.

I came across a really good back-testing vid recently, will try and dig it out and post it.

[eightbo]

adding to this, need to work around your own problems / trading personality based on your journalling or patterns you notice in yourself.

general advice like "Manage emotions" and "create a risk management plan" are too vague in my experience, you need to really create something bespoke which works for you which includes finer details.

as an example, risk management was always pain in my arse trading manually. knowing i needed to fix it did nothing, I felt like there was no solution for a long time and kept up carrying over pre-race trades in play on the racing and freezing up, watching large chunks of my bank at risk unplanned.
I noticed I couldn't close trades even on small risk sometimes due to discipline, but my entries were generally pretty decent, so switching to value betting and dropping size so direct hits can be easily stomached proved a good move for me personally, and EV probably went up too due to not forcing exits just to reduce liability, as well as reduced stress levels. so you could say I'm more of a "gambler" now than a "trader".
weirdly i also noticed trading commodities etc. presented minimal issues as well — place entry and stop and walk away couldn't trigger emotional responses like watching the market tick up and down did. you just have to find what works for you and do that, whatever you're trading / gambling on

[firlandsfarm]

Back testing on real tick data has been the strongest confirmation to sticking to these plans for me.

I came across a really good back-testing vid recently, will try and dig it out and post it.

Here's the video. The first minute sets the scene (yes it's from financial market scenarios but Peter makes the same conclusions in his recent video). It's all about random data and yes I believe sports results are random. One outcome may be more likely but whether the likelihood of an outcome is the outcome is random. The first 'light bulb' moment comes at 4 minutes but that's not to say you should skip the intervening 3 minutes, they are to draw you in! I would recommend all 'back-testers' watch the first half of the video (14 mins.). (The second half is about a system the guy claims is the bees-knees. Maybe it is and maybe it isn't and I don't see it has much application with sports trading but that shouldn't detract from the first half.)

There may be reasons why the first half of the video doesn't apply to someone's back testing methods but it may cause them to think. I back test and will continue to do so but with a maybe more sceptical view of the results and more variation to the parameters.

[ShaunWhite]

Never back test on the same data you used to develop the strategy.

[jimibt]

Never back test on the same data you used to develop the strategy.

yup -80/20 rule!! and also try variations on whether the 20% [testing set] comes before or after (tho preferably after) your 80% [training set] that is back tested against.

I personally back test using this philosophy, usually starting with the previous year split. This is segmented into the 80/20 split and trained against the 80%. I then run these paramaters against the unseen 20% test data. Rinse/repeat until acceptable outcomes.

I then run this across a 3 year full ytd period to see how it fares and until the properties such as profit factor, max equity drawdown, sharpe ratio etc all fall within practical acceptance zones. can be a juggle and is one that needs complete patience. i also pay attention to the fact that one optimisation can seem like a complete golden ticket until you zoom in and see that it produces a substantially lower number of trades. whilst not in itself a BAD thing, i prefer to have a critical mass of trade volume to support the parameters.

[edit] - for clarity, asked copilot to summerise

The 80/20 rule in backtesting, also known as the training/testing split, is a common practice in data analysis and machine learning. Here's how it works:

Training Set (80%): You use 80% of your historical data to train or optimize your trading strategy. This involves adjusting your strategy parameters to fit this subset of data as accurately as possible. The goal is to find patterns and relationships that can be exploited for profitable trading.

Testing Set (20%): The remaining 20% of the data is set aside as the testing set. This data is not used during the training phase. Once your strategy is optimized on the training set, you test it on this unseen data to evaluate its performance. This helps to ensure that your strategy is not just overfitting to the training data but can also perform well on new, unseen data.

Why Use the 80/20 Rule?
Avoid Overfitting: By optimizing on only 80% of the data, you reduce the risk of overfitting, where your strategy might perform exceptionally well on the training data but poorly on new data.
Validation: Testing on the unseen 20% provides a more realistic assessment of how your strategy might perform in real-world trading.
Robustness: A strategy that performs well on both the training and testing sets is likely to be more robust and reliable.

Steps to Implement the 80/20 Rule
Split Your Data: Divide your historical data into two sets: 80% for training and 20% for testing.
Optimize on Training Set: Adjust your strategy parameters using the training set.
Test on Testing Set: Apply your optimized strategy to the testing set and evaluate its performance.
Analyze Results: Compare the performance metrics from both sets to ensure consistency and robustness.

[firlandsfarm]

Considering some of the points made in the video you should maybe run your 80/20 back test many times with different 80 and 20 to check for consistent results. I usually back test 50/50. We don't have the luxury of being able to sample different action times as you can when dealing with financial instruments (unless trading a time based strategy) but we can try different periods. At the end of the day there is an element of 'fitting' with the 80 data and then seeing if the 'fit' also works with the 20 data.

[jimibt]

Considering some of the points made in the video you should maybe run your 80/20 back test many times with different 80 and 20 to check for consistent results. I usually back test 50/50. We don't have the luxury of being able to sample different action times as you can when dealing with financial instruments (unless trading a time based strategy) but we can try different periods. At the end of the day there is an element of 'fitting' with the 80 data and then seeing if the 'fit' also works with the 20 data.

i agree, it's a pragmatic middle ground. 80/20 works for me and have been using it for 3 years now (whether that be continual adjustment or new approaches). one guy that i respect in the space did comment that altho the market trends (zoomed out) look the same, the intraday is vastly evolving and thus maybe 12 months of optimisation is less of a representation than 90 days (train/test ratios may need to be refined for shorter periods maybe).

[Mike Oxlong]

Yes, as the previous comment stated, cross validation is the solution to back-testing / training any models in additional to what Jimbit said. You should try to avoid data leakage by only training on data that would not be available at the given time (training on future runs to predict runs that had occurred prior).

Interesting topic for conversation.

See:
https://en.m.wikipedia.org/wiki/Cross-v ... tatistics)

https://en.m.wikipedia.org/wiki/Leakage ... _learning)

[Michael5482]

I spent loads of time backtesting it was all smoke and mirrors. I have two prices my price and the price the market is offering me, my P&L tells me the rest.

Each to their own though, more than one way to skin a cat.

[ShaunWhite]

Considering some of the points made in the video you should maybe run your 80/20 back test many times with different 80 and 20 to check for consistent results. I usually back test 50/50. We don't have the luxury of being able to sample different action times as you can when dealing with financial instruments (unless trading a time based strategy) but we can try different periods. At the end of the day there is an element of 'fitting' with the 80 data and then seeing if the 'fit' also works with the 20 data.

i agree, it's a pragmatic middle ground. 80/20 works for me and have been using it for 3 years now (whether that be continual adjustment or new approaches). one guy that i respect in the space did comment that altho the market trends (zoomed out) look the same, the intraday is vastly evolving and thus maybe 12 months of optimisation is less of a representation than 90 days (train/test ratios may need to be refined for shorter periods maybe).

I do periodic re-randomisation, but probably more like 20% and lots of other 20%s because I've got over 6Tb of price data so doing 50/50 would take weeks.

[jimibt]

Considering some of the points made in the video you should maybe run your 80/20 back test many times with different 80 and 20 to check for consistent results. I usually back test 50/50. We don't have the luxury of being able to sample different action times as you can when dealing with financial instruments (unless trading a time based strategy) but we can try different periods. At the end of the day there is an element of 'fitting' with the 80 data and then seeing if the 'fit' also works with the 20 data.

i agree, it's a pragmatic middle ground. 80/20 works for me and have been using it for 3 years now (whether that be continual adjustment or new approaches). one guy that i respect in the space did comment that altho the market trends (zoomed out) look the same, the intraday is vastly evolving and thus maybe 12 months of optimisation is less of a representation than 90 days (train/test ratios may need to be refined for shorter periods maybe).

I do periodic re-randomisation, but probably more like 20% and lots of other 20%s because I've got over 6Tb of price data so doing 50/50 would take weeks.

feel your pain - i use ctrader and its optimisation engine. it runs parallel threads across the different parameter sets and on a fast machine with decent memory can churn thro a good few years of data across 1000's of simulations in a short space of time (depending on the complexity of the code in the algo rule). if you have any interest in this area, i think you'd love it as it's a c# interface where you can go as tame or as wild as you like.

anyway, know you're committed on other stuff, but happy to run it by you at some point.

[ShaunWhite]

feel your pain - i use ctrader and its optimisation engine. it runs parallel threads across the different parameter sets and on a fast machine with decent memory can churn thro a good few years of data across 1000's of simulations in a short space of time (depending on the complexity of the code in the algo rule). if you have any interest in this area, i think you'd love it as it's a c# interface where you can go as tame or as wild as you like.

anyway, know you're committed on other stuff, but happy to run it by you at some point.

Thanks Jim. Sounds interesting and similar ish to what I'm doing. But I'm having a big methodology rethink lately so torn about where to direct the effort. ML is getting better and gpt has made using it easier (glad I didn't waste too much time trying to learn it myself). Plus I'm trying to move more towards football, about a year of data so far, and that's going to mean rethinking my approach if what I do now isn't transferable. Ie I've no idea whatsoever what Dec 2025 will look like.

Whoda though having a bet would be so hard

[jimibt]

feel your pain - i use ctrader and its optimisation engine. it runs parallel threads across the different parameter sets and on a fast machine with decent memory can churn thro a good few years of data across 1000's of simulations in a short space of time (depending on the complexity of the code in the algo rule). if you have any interest in this area, i think you'd love it as it's a c# interface where you can go as tame or as wild as you like.

anyway, know you're committed on other stuff, but happy to run it by you at some point.

Thanks Jim. Sounds interesting and similar ish to what I'm doing. But I'm having a big methodology rethink lately so torn about where to direct the effort. ML is getting better and gpt has made using it easier (glad I didn't waste too much time trying to learn it myself). Plus I'm trying to move more towards football, about a year of data so far, and that's going to mean rethinking my approach if what I do now isn't transferable. Ie I've no idea whatsoever what Dec 2025 will look like.

Whoda though having a bet would be so hard

you bet... ooo -get me coat!!

talk in detail when the time works for both