Beta Momentum V3

[GlennConti]

Hello All.
I major league crashed and burned Beta Momentum V1 after a great start. Well I learned a thing or two from that experience. I would like to invite members to simulate the New and Improved Beta Momentum V3. Trust is earned. I am not asking for subscribers, but to just put this strategy on your radar. I am going for Trades Own Strategy (TOS) after I get enough trades.

One of the reasons BM V1 crashed and burned is that I needed $25,000 minimum to day trade the strategy. With such a large capital requirement, my rate of return was too low for my tastes. I took on too much risk chasing returns.

The new Beta Momentum V3 is instructive to options traders of Level 2 that have cash accounts. I am only placing 1 or 2 trades per day so the capital requirement is only about $2,500; one tenth of what I was doing prior. In fact I have funded my IBKR account with less than $1,000. The target is to make $50 profit per day. Thereby I take on substantially less risk and have huge return percentages.

I am expecting a long time period of monitoring my new strategy before people are willing to risk subscribing. Again, trust is earned. But I keep trading my own money and am in this for the long haul.

Personally, I am a subscriber in addition to being a manager and will scale at 200 or 300% soon. Also, I have $1000 in the management driving account and $1500 in the investor target account. I funded this $2500 this year and they are both Roth IRA accounts. I am 65 years old so any earnings on Roth accounts older than 5 years are tax free.

Well that’s about it. Talk is cheap. Show it. Thanks for getting this far.

—Glenn Conti

[GlennConti]

Beta Momentum V3 - “New and Improved Over V1”

[JITF]

I’ll wait till most interesting part - calculating of the future profits of the subscribers. Only then I am going to grab some .

[GlennConti]

Well JITF,

I don’t want to disappoint. Here goes the current prediction as of Friday 01/12/2024…

We are 23 days into the new and improved Beta Momentum V3 strategy. Here is the current prediction for future profits:

75% winners at $45 per win.
25% losers at $92 per loss.
20 trading days per month.

This implies that on a typical day the strategy would make $10.75 per day or $215 per month
[(20 * .75 * $45) - (20 * .25 * $92)]/20 = $10.75
$10.75 * 20 = $215

Current capital at risk = $1043
So, [($215/$1043) * 100] = 20.6% return per month.

What do you think?

[GlennConti]

I’m liking what I’m seeing so far. The driving account has about $1000 in it and I have 2 subscriber accounts capable of 800% additional scaling (500% max+ 300%max). By 1/19/2024, I will be funded in the IB accounts and AutoTrading at full scaling. Total funding about $6500 and a predicted profit of 9 * $10.75 or $97 per day.

–Cheers

My #1 job now is to drop the average loss down from $92 to about $70. By doing this, average daily profit will go from $10.75 per day to $16.25 per day. Plus #2 job, I believe I can drive the winning days percentage up from 75% to something higher. Small changes in my profit take dollar number lower can have a big affect on winning days percentage because I am not sitting around in a trade as long waiting to hit a higher profit number. Let’s see?

[GlennConti]

So, to be clear what I’m doing and how you too can easily predict future profits (hypothetical).

1. I will make just one round trip trade per day.
2. Profit take will be $45. You see this from profit per trade figure on the strategy statistics page.
3. Loss take will be $70. Again, you can see this from the loss per trade figure on the strategy statistics page.
4. The percentage of wins from the statistics page is readily available.
5. The typical calendar month has 20 trading days.
6. The options contract I select will usually have a value at risk of $400 for any given day. If you fund your brokerage account/virtual trading account with $1000, you will be able to autotrade even with an extended string of losing days (targeted at $70). Currently, only 25% of the days will be losers, or about 1 or 2 a week.
7. I am guessing the direction of the trade at market open. Right now I’m correct 75% of the time and hit my profit take in just a few minutes. The less time I am in a trade, the greater the chance that momentum won’t change from news or a whale buying or selling. The longer I am in a trade, the more out of control the situation and the risk. Look to the strategy statistic “Avg Trade Duration”. Right now it’s about an hour. Look for that number to come down.

Well that’s the full disclosure. You can now predict the future profits (hypothetical) yourself. Easy peasy!(1)

(1) Note: Currently(1/13/24) there are only 12 data points (trading days) calculating the averages in the strategy (Beta Momentum V3) statistics. The more data points there are the more stable the predictions become. You will see a large standard deviation in the monthly profit predictions from day to day probability until there are about 100(2) observations (trading days). For example, if you use the prediction model today after 12 trades, 9 wins/12 total trades is 75%. One additional day of a loss would be 9/13 or 69.2%. One additional day of a win would be 10/13 or 76.9%. Total range is 7.7% (76.9 minus 69.2). Also, when you scale you are multiplying the range or variance or error. If you use the prediction model but wait 100 trading days it takes us out to 5/6/2024. The data provided by C2 about the strategy will be much better and wild swings will/should subside by then. Consider this, 75/100 is 75% so one additional win or loss after 100 days is 74.3% (75/101) for a loss and 75.2% (76/101) for a win. That is, after 100 days one additional win or loss has much less effect on the win percentage. Total range is .9% (75.2 minus 74.3) so there is much less error. However, you do have to wait 100 days to get greater certainty or less error, but If you are less conservative, wait 50 days until 3/7/2024.

[GlennConti]

Beta Momentum V3 Running Profit Projections XLSX

I will endeavor to update this spreadsheet each trading day from C2 statistics data for Beta Momentum V3 strategy.

[GlennConti]

Please also consider this. I just asked the AI from Microsoft Copilot how many free throws you need to see from a player to determine his or her free throw percentage. Here is the answer: 16.

That’s an interesting question. To measure a basketball player’s free throw percentage, you need to divide the number of free throws made by the number of free throws attempted12. However, the number of shots or measurements needed to get an accurate gauge of the player’s performance depends on several factors, such as:

• The level of confidence and precision you want in your estimate
• The expected variability or range of the player’s free throw percentage
• The available time and resources for collecting and analyzing the data

One possible way to determine the sample size is to use a statistical method called power analysis23. This method allows you to calculate the minimum number of shots or measurements required to detect a certain effect size (such as the difference between two players’ free throw percentages) with a given level of confidence and power. Power analysis can also help you avoid collecting too much or too little data, which can lead to inaccurate or inconclusive results.

To perform a power analysis, you need to specify some parameters, such as:

• The significance level (alpha), which is the probability of rejecting the null hypothesis when it is true (usually set at 0.05 or 5%)
• The power (1 - beta), which is the probability of detecting the effect when it is present (usually set at 0.8 or 80%)
• The effect size (d), which is the standardized measure of the difference between two groups or conditions (usually calculated as the mean difference divided by the standard deviation)
• The population standard deviation (sigma), which is the measure of the variability or spread of the data (usually estimated from previous studies or pilot data)

Once you have these parameters, you can use a formula or a software tool to calculate the sample size (n) for your study. For example, using the formula for a two-sample t-test, the sample size is given by:

n=2⋅(Zα/2+Zβd)2⋅σ2n=2⋅(dZα/2​+Zβ​​)2⋅σ2

where Zα/2Zα/2​ and ZβZβ​ are the critical values of the standard normal distribution corresponding to the significance level and the power, respectively.

As an illustration, suppose you want to compare the free throw percentages of two players, A and B, with a significance level of 0.05 and a power of 0.8. You expect that player A has a free throw percentage of 80%, while player B has a free throw percentage of 70%. You also estimate that the population standard deviation is 10%. To calculate the effect size, you can use the formula:

d=μA−μBσ=0.8−0.70.1=1d=σμA​−μB​​=0.10.8−0.7​=1

where μAμA​ and μBμB​ are the population means of the free throw percentages of players A and B, respectively.

Using a software tool or a table, you can find that the critical values of the standard normal distribution are Zα/2=1.96Zα/2​=1.96 and Zβ=0.84Zβ​=0.84. Plugging these values into the formula, you get:

n=2⋅(1.96+0.841)2⋅0.12=15.5n=2⋅(11.96+0.84​)2⋅0.12=15.5

Since you cannot have a fractional number of shots or measurements, you need to round up the sample size to the next integer, which is 16. This means that you need to observe each player taking 16 free throws to get an accurate gauge of their performance.

Of course, this is just one example of how to determine the sample size for measuring free throw percentage. Depending on your research question and design, you may need to use a different method or formula. You can also adjust the parameters to suit your needs and preferences. For instance, if you want a higher level of confidence or power, you will need a larger sample size. Conversely, if you want a smaller effect size or a lower variability, you will need a smaller sample size.

I hope this helps you understand how to measure a human’s free throw percentage in basketball and how to choose an appropriate sample size for your data set. If you have any further questions, please feel free to ask me.

[GlennConti]

So, possibly measuring my ability to win a trade only takes 16 days at one trade per day.

[GlennConti]

Here is Google’s AI Bard’s answer:

Desired level of accuracy:

• Low accuracy: Just a handful of shots (5-10) can give a rough estimate, but can fluctuate wildly with small changes in performance.
• Moderate accuracy: 20-50 shots might offer a more reliable picture, but may still be susceptible to random variations.
• High accuracy: For statistically robust estimates, you’d need a larger sample size,typically at least 100 or even 200 shots. This helps iron out small fluctuations and provides a more reliable indicator of a player’s true free throw percentage.

Player experience and consistency:

• Seasoned players with consistent form: Fewer shots might be sufficient to establish their free throw percentage due to their established skill and consistent mechanics.
• Rookies or inconsistent players: More shots are needed to account for potential fluctuations in their performance and identify underlying trends.

[InteractiveAssets]

I think that isn’t reasonable.

Edit: specifically this “ So, possibly measuring my ability to win a trade only takes 16 days at one trade per day.”

[GlennConti]

If anything it is well reasoned or over reasoned. It’s all about determining my correct free throw shot percentage or ability to correctly determine the market price trend. Right now C2 data says I can do it 75 percent of the time. However, that is only based on 12 trades or data points. If after 200 trades or data points my winning trade percentage is still 75%, the statistical science says that number as a predictor is pretty good.

[GlennConti]

So, time will tell what is my winning trade percentage. Copilot AI says it can be known accurately with as few as 16 trades. Bard AI says medium confidence at 20 to 50 trades and high confidence at 100 trades.

Once we know my winning trade percentage(WTP) accurately and we know how much $ I win per win(DPW) and how much $ I lose per loss(DPL), predictions are simple math.

Predicted Yield Per Trade(PYPT) is simple math.

PYPT = ((WTP * DPW) - ((1-WTP) * DPL))

Yes, totally reasoned and reasonable.

What is incredible is a 75% WTP. If I just determined the trade direction at random, it would be 50%. 75% is significantly higher performance on my part than chance. Is that…

1. A fluke that will go lower as we get more data
2. Proof that I have trading skills
3. Proof that C2 data collection and tabulation is hypothetical

???

I am tracking the PYPT and WTP and DPW and DPL
In a spreadsheet here…

Beta Momentum V3

Beta Momentum V3 Running Profit Projections XLSX I will endeavor to update this spreadsheet each trading day from C2 statistics data for Beta Momentum V3 strategy.

[GlennConti]

I just had Copilot AI build me the above table. As the sample size (number of trades) increases from 10 to 100, the margin of error goes down. Therefore, the predictive capacity of the winning trade percentage goes up.

Here is a log table:

[GlennConti]

So, if you want a formula for Margin of Error as a function of Sample Size it is as follows:

Y = 1.12/(x**0.531)

Or in Excel

=(1.12/(POWER(x,0.531)))

[GlennConti]

BOTTOM LINE:

After 12 trades as a sample the margin of error according to Artificial Intelligence is 29.93%

[GlennConti]

More comedy as far as predictions for future results.

This spreadsheet is available at the above link.

[GlennConti]

So, I just asked Copilot AI:

if I have an 80% chance of winning $45 on a win of a coin flip bet and 20% chance of losing $90 on a loss of the bet, should I take the bet?

AI says:

To answer this question, we can calculate the expected value of the bet, which is the average outcome of the bet over many trials. The expected value is given by the formula:

$$\text{expected value} = \text{win probability} \times \text{win amount} - \text{lose probability} \times \text{lose amount}$$

Using the values given in your question, we get:

$$\text{expected value} = 0.8 \times 45 - 0.2 \times 90 = 18$$

This means that on average, you will gain $18 for every bet you make. Therefore, it seems like a good bet to take, as long as you are comfortable with the risk of losing $90 in some cases.

I used the Python code below to calculate the expected value, which you can also run and modify if you want.

def expected_value(win_prob, win_amount, lose_prob, lose_amount):
    # win_prob is the probability of winning the bet
    # win_amount is the amount of money gained if the bet is won
    # lose_prob is the probability of losing the bet
    # lose_amount is the amount of money lost if the bet is lost
    # expected value is the average outcome of the bet over many trials
    return win_prob * win_amount - lose_prob * lose_amount

win_prob = 0.8 # 80% chance of winning
win_amount = 45 # $45 gained if the bet is won
lose_prob = 0.2 # 20% chance of losing
lose_amount = 90 # $90 lost if the bet is lost
print(f"The expected value of the bet of a coin flip is {expected_value(win_prob, win_amount, lose_prob, lose_amount)}.")

The expected value of the bet of a coin flip is 18.0.

I hope this answers your question. If you have any other questions or requests, please let me know.

Caveat: The “coin flip” in this case is me properly predicting the market trend in the next near future either up (heads) or down (tails). The odds 80% winners and 20% losers have been established after only 15 trials. The margin of error is approximately 26.5% because there needs to be more accuracy (more trading days) to establish a higher confidence level. Look to February 1st (25 trades/samples will be history) and March 9th (50 trades/samples will be history).

Take this for what it is worth.

[GlennConti]

After 15 days (samples) 80% winners and 20% losers is my track record today.

[GlennConti]

This augmented stat sheet for Beta Momentum V3 keeps evolving. Note the new table of probabilities for long win or loss strings.

Current winning trades percent: 81.25%
Current win streak: 8 days.