BEST MOVING AVERAGES HQ:
For Short-Term Trading (Scalping / Day Trading)
1.9 EMA (Exponential) โ reacts fast, great for quick entries/exits.
2.20 EMA โ balance between noise & trend, common intraday guide.
๐น For Swing Trading:
3.50 SMA (Simple) โ classic trend filter, used by institutions.
4.21 EMA โ smooths short swings without too much lag.
๐น For Long-Term / Position Trading:
5.100 SMA โ strong trend indicator on daily/weekly charts.
6.200 SMA โ the king of long-term trend direction; above = bullish, below = bearish.
Donโt just use one. Combine them.
Example: 20 EMA + 50 SMA + 200 SMA โ short, medium, long-term trend alignment.
The @the_nof1 DeepSeek Model is already outperforming 99% of Traders.
HyperScore of 6.1 out of 10
Sharp Ratio of 13.3
Max DD of 15.52%
Why I like it?
It's automated. Rather than discretionary Vault Strategies I don't have to fear the usual outperformance of 3 weeks to be followed by a total wipe out because of poor risk management.
You can easily track and copy trade it on Tools like @HyperSignals_ai
Wallet
0xc20ac4dc4188660cbf555448af52694ca62b0734
@Alkan1459903 @quantbeckman Don't be disrespectful, man.
If you can't get anything useful from it, just scroll to the next post on your TL.
Don't expect to be spoon-fed success.
@scalpinjimmy@TradeZella When you say 'tested on the entire 2025...', is that an out-of-sample test?
From what you've said, it might not be very wise to trust the results you're observing just yet
There are issues that, although they seem like common sense, are only common in the minds of some #quants. For example, why are firms hiring PhDs just to identify which methods are most appropriate to use with certain data?
One of the main blunders I've encountered time and again is this: Failure to align the modeling method with the #trading modelโs objective. When you use methods created by third parties, the usual result is poor performance and losses. This happens because methods are designed to address specific problems and have inherent limitations that, if ignored, prevent the model from fulfilling its purpose.
@MarginCall4 1 year of 5-minute data still represents just 1 year of market period as 1 year of daily data, but with exponentially more noise and volatility.
If your strategy can extract meaningful signals from such a low signal-to-noise environment, then by all means, go for it.
@MarginCall4 This only works under the assumption that the relationship between more data (derived from more granularity) and statistical significance (or useful information) is linear; which is critically flawed.
@guy_nale You're doing very well, brother. You might not be where you want to be yet, but you're right you need to be. Keep pushing, man.
(PS: you still followed it with the motivational statement, and I dig it๐๐ค)
@MoDivine1@SwarmQuant @TradetroniX Funny enough, I haven't tried this๐ค
So far, in my experiments, I have only used Minmax normalisation (and sigmoid, sometimes) for the distance-related conditions
@TradetroniX @SwarmQuant Basically. You could also combine that with other signals to 'penalize' overbought/oversold conditions.
In a way, it's still the same thing as 'filtering' with extra conditions in a discrete system, but with continuous signals, rather than on/off switches
1/N
Last week, @JoachimMo1985 and @thetradler initiated the "continuous vs binary" conversation, which I found to be very interesting, including all the insights made by other respectable accounts.
I couldn't say anything because I had never tried it out, so I decided to...
I can't tell if this post is satire, or intentionally vague.
It sounds like great advice (or even an edge) until you have to define:
- uptrend / downtrend
- pullback / rallies
- actual vs false breakouts/breakdowns
- support / resistance
- where the 'range' limits are
Five basic trading strategies:
1. Buying pullbacks in an uptrend.
2. Shorting rallies in a downtrend.
3. Buying breakouts in bull markets.
4. Shorting range breakdowns in bear markets.
5. Buying support and shorting resistance in a rangebound market.
These are all valid ways to make money in any market if you manage your risk-to-reward ratio at entry.
@TradetroniX @JoachimMo1985@thetradler I spoke about the image from post 17 of the thread. I agree, it's a bit too lengthy ๐
Those results are from further experiments with it. Particularly, applying weights to each member of the voting system and optimizing those weights on historical data.
@TradetroniX @JoachimMo1985@thetradler That's very interesting. The whole conversation was very insightful. For all my years in experimenting with algorithms, I never tried it. I have also never seen a strategy work as well as this one did "right out of the box" without tweaking parameters.