The Quantist

[locobobo]

I'm going to use this thread to discuss a few of my findings based solely on market modeling, statistics, and related projections, or collectively, quantitative market analysis. If some substance comes across as abstruse, or just plain weird, I ask that you post your queries or concerns.

[locobobo]

Hello everybody,
I'm very new into quant field (not new to classic trading) and that I have a question: Quant. Trading is simply for HF or it's likewise employed for longer periods (days,weeks,months,etc)? Just curious. It may be used for the two - provided you have sufficient data quantity and diversity. Markets are fractal as is their cyclical character.
But, from personal experience, I found most success with briefer term projections since they have a lower probability of encountering shockwaves (big news, reports).

[jaime.camgenas176]

Hi everyone,
I am very new into quant area (not new to classic trading) and that I have a question: Quant. Trading is only for HF or it is likewise used for longer terms(days,weeks,months,etc)? Just curious. It is used for all conditions you may imagine including sampling markets with variable frequency. For example professionally I am accountable for systems that sample and make trading decisions after every few weeks. Talking about extremes the maximum holding period of a single position in my office was 3 years and 5 months

[MigwlPmocmaslinapkm]

I'm largely an intraday trader as intraday data is detrended, and thus shows greatest cyclical trend. Do you detrend the data? I'm playing with spline smoothing, but for my low frequency system it's not really necessary yet.

Do you utilize market regime in any quantitative way? I have a mean reversion system, but I consider the market regime into account. The detection mode is quite simple, no hidden Markov, SVM or state-space models here. My math is not quite there yet

[oxmlukaoxmlu]

I'm mostly an intraday trader as intraday data is most detrended, and thus shows greatest cyclical trend. I believe your conclusion is about the same with my regular hurst exponent analysis, which shows that the period for optimum hurst exponent is about 16-20 hours for the majority of the pairs.

[giomgaybenalli]

I do not think simple necessarily means simple. I also don't think statistical trading needs to be complied. A lot of individuals talk about changing markets, a few things change, some stay the same. I would argue that's true of the markets. I think when it comes down to this, typical trading is all about exploiting human behaviour, and that I think human behaviour is relatively consistent. That's exactly what has changed during the last few years when hft algo trading has increased it's ugly head, a few patterns which have worked for decades no longer work and they did.

[vpinoxn]

That's exactly what has changed during the past couple of years when hft algo trading has raised it's ugly head, a few patterns that have worked for decades no longer work as well as they did. I would consider that technological change, not a change in human behavior. By behavior, I suggest the slow, general behavior that drives longterm movement. I have no doubt that markets change in many of ways over the years, and egies can lose or gain effectiveness, but I think the overall behavior of the participants will probably remain exploitable, and won't change all that considerably. Perhaps that's only true for more timeframes.

[locobobo]

Can you detrend the information? I'm playing with spline smoothing, but for my low frequency method it is not really necessary yet.

Can you use market regime in any quantitative way? I've a mean reversion system, but I take the market regime into account. The detection mode is quite straightforward, no hidden Markov, SVM or state-space models . My math isn't quite there yet Yes I further detrend intraday information, usually by oscillating it.

Market program. . .yes but I call the process quantitative dissecting as you literally select market waves apart. Main concentrate on frequency domain, and time for significance.

[locobobo]

I think that your conclusion is about the same with my regular hurst exponent analysis, which shows that the interval for optimal hurst exponent is about 16-20 hours for most of the pairs.
Ah superb, the number seems about perfect.

[zomska]

Yes I further detrend intraday information, usually by oscillating it.

Market regime. . .yes but I call the procedure quantitative dissecting because you literally select market waves apart. Main focus on time for correlation name, and frequency domain. Could you point to some source references for description of the method you use - if generally merely...?

Thanks

[locobobo]

This was part of a EUR/CAD forecast I printed earlier today.
It's a raw projection with its frequency decomposition, made just before April 13th. Its correlation to price action stands comparatively significant. Again, pay attention to the waveform instead of purely amplitude.
https://www.nigeriaforextrading.com/attachments/15194213201387688920.png

[juggyll]

...
I am largely an intraday trader as intraday data is detrended, and thus shows greatest cyclical tendency. intraday. . Does this mean u trade a specific timeframe? If so, which one/s?
Due