What do you base your speculation on?

[Dansker]

As the subject reads: What do you base your speculations and prediction on the Bitcoin prize on?

Is it a hunch? News? current scandals? Maths and statistics?

[1714947870]

1714947870

[1714947870]

1714947870

[1714947870]

1714947870

[labestiol]

My secret method : Bitcoin is awesome -> price will go up.
Know what ? It works 

[Dansker]

My secret method : Bitcoin is awesome -> price will go up.
Know what ? It works 

You wouldn't make a very good day trader I think

[labestiol]

My secret method : Bitcoin is awesome -> price will go up.
Know what ? It works  

You wouldn't make a very good day trader I think

I agree. I've never been convinced by the risk/rewards of daytrading anyway. Especially daytrading bitcoin.
Just too small of a market, too unpredictable imo.
Would be really curious to see the books of people doing it (for at least months).

[Vandroiy]

First and foremost, maths and statistics. But the interesting question is what to apply them on.

I believe it's best to first apply them to myself: figure out how good my own predictions are, and make sure I'm ready to face my own mistakes. No point in getting cornered for maximum short-term profit. For the actual predictions, I use a combination of simplified models, psychological ideas, and historic comparisons, all fed by whatever indicators I can get a hold of: news, forum stats, various forms of trading volume, ...

The difficult part for me is calibrating a model after I have some rough idea. What's the scale, growth, sentiment of Bitcoin? And I have to cope with horribly foggy indicators. I'm not really versed in history, so I have to look for past events that hold similarities. This can be within Bitcoin for short-term things, or old stuff like the DJIA around 1930 to understand what happened in the big 2011 Bitcoin bubble.

Don't take this too seriously, I'm still a noob. I'm not even done "calibrating" myself, don't know how many mistakes I do on average. So this post could be one.

[Realpra]

Daytrading is just gambling imo.

I'm long Bitcoin because logically it makes sense to me as a monetary system in a world with a very broken monetary system.

Short term people can be irrational, but over 10-20 years logic always wins. Consumerism, inflating fiat and bailouts never made any sense.

[knight22]

I'm basing on humans behaviors confronted to different scales of events.

[kentrolla]

my crystal ball helps a lot

[adamstgBit]

i take in all the  News, current scandals, BS Maths and statistics some poeple post here. and formulate a hunch!

oh and my hunch is that 10$ is the new 5

my fear is the 15 is the new 32 

My fears are slowly going away

[ElectricMucus]

i take in all the  News, current scandals, BS Maths and statistics some poeple post here. and formulate a hunch!

oh and my hunch is that 10$ is the new 5

my fear is the 15 is the new 32 

My fears are slowly going away

They can both be true...

[beckspace]

Humans...

11.11 is the new 6.66

[ElectricMucus]

Multi-variant empirical Bayesian recursion. Model risk parameters chosen through a simulated annealing process.

I am still waiting for the day you share anything more substantial than that

[adamstgBit]

i take in all the  News, current scandals, BS Maths and statistics some poeple post here. and formulate a hunch!

oh and my hunch is that 10$ is the new 5

my fear is the 15 is the new 32  

My fears are slowly going away

They can both be true...

after a few million coins have changed hands at around 10$ i think its safe to assume 10 is the new 5

so far so good.

[ElectricMucus]

Multi-variant empirical Bayesian recursion. Model risk parameters chosen through a simulated annealing process.

I am still waiting for the day you share anything more substantial than that

Bayesian recursion is something that is understood, I'm not inventing anything here. I am using empirical Bayes because I am not smart enough to choose a prior that works better than an empirical curve. The eight variants that I am using now come from a constant volume series. That is, I first redimension the time series into candles of similar volume. That bit is not hard, although, sometimes when I make a change to the model I have a hard time figuring out how to get the posterior to converge. The hard part is figuring out where to set risk parameters, that is the maximum amount of available liquidity to place in any given order. Too little and profits are lost from not enough trade volume, too much and the risk is liquidity. Balancing skewness and kurtosis risk is tricky and the search space for the model is large. It is probably either an NP-complete or NP-hard problem actually. The best way I have discovered so far to find solutions that meet my fitness criteria, within my computing budget, is a form of simulated annealing. This is also something that is understood, I am not inventing anything.

Thanks, I don't quite understand what you are talking about but that may be the point
Have you tried modeling it after a gamma distribution?

Can you say for your model to be back-tested in a rigorous way based upon available past data or is there still some part of it prone to subjectivity?

[tiberiandusk]

[julz]

I base mine on gut feeling about bitcoin news and scandals.

This has been unsuccessful for me.
I only trade with pretty small amounts, and I made about 2 bitcoins last month doing the occasional day trade - but it wasn't worth the time and stress.


Now I'm holding some USD when I'd prefer to be all BTC.  If the price doesn't drop soon then I guess I'll buy back in at a loss and wipe out my previous little BTC gains plus some. :/


Basically - I suck at day trading, but every now and then I forget that and get sucked in to giving it another whirl.

[tvbcof]

I base my speculation almost completely on my own theory of the value of such a solution (I think.)

Associated factors include

 - the likelihood of failure or supplantation of the solution.
 - the possible reactions (namely corp/gov) to the solution.
 - the potential trajectories of Bitcoin which could provoke failure or success.
 - the potential trajectories of fiat currency solutions which could provoke failure or success.

Like most of my speculative adventures, I'm not anticipating success for years (if at all) so I don't pay much attention to the routine things which pop up (e.g., thefts, scams, tv shows, etc.)  At least not for the purpose of modulating the magnitude of my speculative investment.  Some of them do lend (or reduce) strength the the various hypothesis I hold about factors effecting the chances of a particular outcome.  And a lot of these routine things are fairly amusing.

---

A completely unrelated factor in my interest in Bitcoin is political.  It is a good way to actively demonstrate my rejection of the USD, for instance.

[labestiol]

Multi-variant empirical Bayesian recursion. Model risk parameters chosen through a simulated annealing process.

I am still waiting for the day you share anything more substantial than that

Bayesian recursion is something that is understood, I'm not inventing anything here. I am using empirical Bayes because I am not smart enough to choose a prior that works better than an empirical curve. The eight variants that I am using now come from a constant volume series. That is, I first redimension the time series into candles of similar volume. That bit is not hard, although, sometimes when I make a change to the model I have a hard time figuring out how to get the posterior to converge. The hard part is figuring out where to set risk parameters, that is the maximum amount of available liquidity to place in any given order. Too little and profits are lost from not enough trade volume, too much and the risk is liquidity. Balancing skewness and kurtosis risk is tricky and the search space for the model is large. It is probably either an NP-complete or NP-hard problem actually. The best way I have discovered so far to find solutions that meet my fitness criteria, within my computing budget, is a form of simulated annealing. This is also something that is understood, I am not inventing anything.

Thanks, I don't quite understand what you are talking about but that may be the point

Can you say for your model to be back-tested in a rigorous way based upon available past data or is there still some part of it prone to subjectivity?

Specifically, I back test over the last 400 constant volume candles, right now that means that each candle is for price activity over a span of about 14k BTC of volume. The data used for backtesting goes back about 100 days, I don't actually find it useful to go back any farther.

As someone who build models for a living (system biology), I like what you're talking about.
Still, my question is : Does it work ?

[labestiol]

Multi-variant empirical Bayesian recursion. Model risk parameters chosen through a simulated annealing process.

I am still waiting for the day you share anything more substantial than that

Bayesian recursion is something that is understood, I'm not inventing anything here. I am using empirical Bayes because I am not smart enough to choose a prior that works better than an empirical curve. The eight variants that I am using now come from a constant volume series. That is, I first redimension the time series into candles of similar volume. That bit is not hard, although, sometimes when I make a change to the model I have a hard time figuring out how to get the posterior to converge. The hard part is figuring out where to set risk parameters, that is the maximum amount of available liquidity to place in any given order. Too little and profits are lost from not enough trade volume, too much and the risk is liquidity. Balancing skewness and kurtosis risk is tricky and the search space for the model is large. It is probably either an NP-complete or NP-hard problem actually. The best way I have discovered so far to find solutions that meet my fitness criteria, within my computing budget, is a form of simulated annealing. This is also something that is understood, I am not inventing anything.

Thanks, I don't quite understand what you are talking about but that may be the point

Can you say for your model to be back-tested in a rigorous way based upon available past data or is there still some part of it prone to subjectivity?

Specifically, I back test over the last 400 constant volume candles, right now that means that each candle is for price activity over a span of about 14k BTC of volume. The data used for backtesting goes back about 100 days, I don't actually find it useful to go back any farther.

As someone who build models for a living (system biology), I like what you're talking about.
Still, my question is : Does it work ?

It beats the market in the time period that I test in, that is my fitness criteria. Even despite spotty execution I have more BTC than I started with every month.

Do you think your time period is significant enough ?
I really need to take some time off to play with that 

[bitcoinBull]

It beats the market in the time period that I test in, that is my fitness criteria. Even despite spotty execution I have more BTC than I started with every month.

What time range for holding periods does your model calculate?