Almost everyone is lying. Because there's no such...

[ZTrader888]

I agree with your post. Mathematics is just a tool. The reality of financial markets is that it's impossible to incorporate all the variables. Just do a search in Google Scholar for tensor analysis of financial markets and the problems inherent in analyzing financial markets becomes apparent.

The post below appeared on Medium. The author gives some good guidance as to how we should be looking at financial markets and the models we build around them.

Mathematics is never out of the picture - but must be tooled to suit the problem. Quantum mechanics is a classic example where probability theory replaced classical mathematics to describe the quantum world. In financial markets, there are limits as to what we can know at any one point in time regardless how much data we analyze. It's the nature of the beast.

The Trinity Of Errors In Financial Models: An Introductory Analysis Using TensorFlow Probability

"
Finance is not a precise predictive science like physics. Not even close. So let’s not treat academic theories and models of finance as if they were models of quantum physics.

All financial models, whether based on academic theories or data mining strategies, are at the mercy of the trinity of modeling errors. While this trifecta of errors can be mitigated with appropriate modeling tools, it cannot be eliminated. There will always be asymmetry of information and cognitive biases. Models of asset values and risks will change over time due to the dynamic nature of capitalism, human behavior, and technological innovation."

 

[illstep]

What kind of statistical significance would one need to think that he MIGHT have some sort of an edge'?

I'm someone whose been trying to trade for about 4 years and has been consistently losing. Almost everyday. Can't buy a winning day.

Sometimes i'll find what I think might be an 'edge' through lots of backtesting, but when live trading, it tends to break even at best. (on sim and real money)

In fact, in all my time trying to do this, i've never had a 'spell' of LIVE trading (either SIM or real money) that's made me think ''Hmmmm....I wonder if i'm onto something here! This is my 2nd week of making money, adn i've made almost every day!''.

Until very recently that is. I seem to make money most days at the moment. Although i'm super pessimistic (realistic?) by nature about this sort of thing so i'm not getting carried away. At all!! Could always just be luck, I guess

If you are daytrading the futures, and maybe you typically take around 6 trades per day (stops of maybe 2 points and targets maybe slightly bigger, although all dependent on various technicals) what sort of results should give you confidence? How many trades/day trading would you need for a good sample size?

 

[Mnewton]

Forget all the analysis , pick a stock that moves daily and watch it.
you need to trade 1000 shares and make 20 cents. do that every day
Dont trade the afternoon you will lose.
Watch a few stocks from 9.30 to 10.30 and from 10.30 to 12.
pick the one you want to trade. start low at 100 shares and work up.
Do not over trade. build up a trading acount. put some $$ aside for longs.
buy something like a FB at $30 and wait until it hits 200, maybe UBER.
Every day is a new day no back testing will help, looking at the charts
and momentum will. Keep at it.

 

[monet]

The reason nothing worked for you is everything you tried was too simplistic.

To clarify, simplistic is not the same as simple.

A simple profitable system should be the goal, but testing pin bars in isolation without considering a larger context is too simplistic.

 

[JigsawTrading]

Agree and Disagree. Ultimately everything is a math problem and there is nothing that a human trader does that is magic that can't also be done by an algorithm. Difficult yes, but not impossible.

It's true that engineers often fail in markets. But many of the best traders are also engineers. The early failures come because they attempt to apply the classical mathematical tools and techniques that they learned in school, e.g. simple linear regression, which tend to not be suitable for the insanely high noise:signal ratio that exists in markets. The ultimate solution to this problem is to discover appropriate techniques, not to throw up one's hands and say "oh well, it must be more of an art than a science".

OK - if everything is a math problem - tell me how you approach this.

There are 2,500 people in a shopping mall. All at different locations in the mall and it's 3pm in the afternoon -peak shopping time.

Your job is to mathematically predict where each of these 2,500 people will be at 3:15pm.

You will be paid $5 for each correct person placed and will have to pay $5 for each incorrect person.

You do not need to provide the exact math here - but just outline the approach to making a profit if you truly believe EVERYTHING is a math problem.

 

[cruisecontrol]

OK - if everything is a math problem - tell me how you approach this.

There are 2,500 people in a shopping mall. All at different locations in the mall and it's 3pm in the afternoon -peak shopping time.

Your job is to mathematically predict where each of these 2,500 people will be at 3:15pm.

You will be paid $5 for each correct person placed and will have to pay $5 for each incorrect person.

You do not need to provide the exact math here - but just outline the approach to making a profit if you truly believe EVERYTHING is a math problem.

Predicting the exact position of each individual with 100% accuracy would require measurement and computational power that does not exist today and will probably never exist, but is not impossible in theory. If you had complete information about every subatomic particle making up all the people in the mall you could just simulate it and see what the outcome is.

Back to reality, what we can actually do is model probability of various outcomes based on all the data we have. Consider a company like Google or Facebook. Most of the people in that mall are probably carrying smartphones that track their location, so the current location may be known exactly. Additionally, the companies know lots of information about who those people are friends with, so groups can be identified that are likely to stay in proximity. Their interests are known based on stuff they searched for online or what they "liked" online and this will correlate with what shops they are likely to visit. Their past location history is also known - where have they been before in that mall, what shops do they visit repeatedly. Etc, etc. You put all the information together and you can figure out what is more or less likely to happen.

In your scenario there is a reward for correct predictions and loss for incorrect. I just want to note that this isn't a fair test because there is only one way to be correct, but an infinite number of ways to be wrong. Thus even if a model is highly accurate and effective by most metrics, it is unlikely to make money under your conditions.

A more realistic scenario might be: Predict which stores will have the biggest increase in customers in the next 15 minutes and which will have the biggest decrease. Modeling large groups of people can be much more accurate than modeling individuals. Just like in physical sciences there are ways to model the behavior of gasses and liquids in various states, even though the motions of each of the trillions of individual atoms is basically random and unpredictable.

 

[JigsawTrading]

Predicting the exact position of each individual with 100% accuracy would require measurement and computational power that does not exist today and will probably never exist, but is not impossible in theory. If you had complete information about every subatomic particle making up all the people in the mall you could just simulate it and see what the outcome is.

Back to reality, what we can actually do is model probability of various outcomes based on all the data we have. Consider a company like Google or Facebook. Most of the people in that mall are probably carrying smartphones that track their location, so the current location may be known exactly. Additionally, the companies know lots of information about who those people are friends with, so groups can be identified that are likely to stay in proximity. Their interests are known based on stuff they searched for online or what they "liked" online and this will correlate with what shops they are likely to visit. Their past location history is also known - where have they been before in that mall, what shops do they visit repeatedly. Etc, etc. You put all the information together and you can figure out what is more or less likely to happen.

In your scenario there is a reward for correct predictions and loss for incorrect. I just want to note that this isn't a fair test because there is only one way to be correct, but an infinite number of ways to be wrong. Thus even if a model is highly accurate and effective by most metrics, it is unlikely to make money under your conditions.

A more realistic scenario might be: Predict which stores will have the biggest increase in customers in the next 15 minutes and which will have the biggest decrease. Modeling large groups of people can be much more accurate than modeling individuals. Just like in physical sciences there are ways to model the behavior of gasses and liquids in various states, even though the motions of each of the trillions of individual atoms is basically random and unpredictable.

3:10 - Ring fire alarm
3:15 - Everybody is outside

See - everything isn't mathematical

 

[ZTrader888]

OK - if everything is a math problem - tell me how you approach this.

There are 2,500 people in a shopping mall. All at different locations in the mall and it's 3pm in the afternoon -peak shopping time.

Your job is to mathematically predict where each of these 2,500 people will be at 3:15pm.

You will be paid $5 for each correct person placed and will have to pay $5 for each incorrect person.

You do not need to provide the exact math here - but just outline the approach to making a profit if you truly believe EVERYTHING is a math problem.

You're wrong. That's a probability problem. And it's an excellent example because your scalping techniques aren't much different. You're training your clients to calculate the probability of a profit every time they push the button. Your scalping techniques are essentially probability problems. It's up to the trader to work through the problem and learn how to construct some prediction of success. If trading was a totally random process, then all techniques would have an equal chance of success. But we know that's not true. You obviously don't realize it - and this is not uncommon - but you're doing the math in your head all day long when you're trading.

As to solving the 2500 person problem mathematically, there are several ways to approach it. For instance, the law of large numbers, the law of the iterative logarithm, and the central limit theorem. The problem becomes more complex when you have to figure in the variables and the number of tests that must be done to generate a probability distribution. I don't think we want to go there, but you get the idea.

BTW, you have an excellent website - the tools are outstanding.

 

[cruisecontrol]

3:10 - Ring fire alarm
3:15 - Everybody is outside

See - everything isn't mathematical

So you were never interested in the answer, you were just setting me up for what you consider to be a BTFO. Except it really isn't one at all. The probabilty of someone pulling the alarm is just part of the model.

Also,
3:20 - JigsawTrading is arrested for false alarm, caught blue handed.

 

[ZTrader888]

So you were never interested in the answer, you were just setting me up for what you consider to be a BTFO. Except it really isn't one at all. The probabilty of someone pulling the alarm is just part of the model.

Also,
3:20 - JigsawTrading is arrested for false alarm, caught blue handed.

Exactly -