The extra data could be bigger or smaller. Granted, some of the available data is from 1945 on, and some even less, and some from longer than the data that is easy to come by.

No worries. As long as you don't go nuts and get overconfident because you have found something that you find agreeable, you are fine. I will grant you that that is a difficult task.

There is even the whole thing about transportation costs not being a fixed cost. Still, it is all calculable to some extent.

My point is that you can't do much with just knowing that there isn't 100% liquidity. It is like deciding who to date by saying "girls all have a height."

There is very little informational content about where prices will be (which is what I care about) 10 years from in an ECB chairman's speech. The more reasonable things that get people's panties in a bunch are also fairly free of predictive content.

It gets quite expensive when you could be sitting elsewhere. Opportunity costs are costs.

The new data should only sway you a bit. It doesn't have more weight than the old data (unless, of course, you made the initial mistake of over-fitting, but lets assume you haven't).

Of course. I mentioned several techniques a while back that I find useful to check whether I am being silly.

Does the pair seem reasonably likely to have a tendency towards increasing in price? Most things do (because of inflation), but there are notable exceptions.

Okay if by transportation you mean everything including stuff like processing costs then yes its just transportations costs. No doubt it is calculable to and if you can do the calculations you can take advantage of value causationThat's not the same though as just expecting wti and crude to mean revert, its the adjusted figured that mean revert.

I think there is some confusion here. We are not trying to measure changes in liquidity, we are making use of the fact that the lack of 100% liquidity causes prices to move from their value, and that is measurable.

Depends what he says but for those of us concerned with shorter-term values there can be a lot of content. In the long enough term very little matters.

That really is small change. In no way could that be considered expensive.


That really depends on what the new data.

but that is a very weak check It suffers from all sorts of bias that however hard we try we cannot really overcome You are relying on being unrealistically sensible in checking to see if you are being silly!!!

New evidence is what you really need and its nothing to do with frequentism.

I think it may happen to be a property of your particular strategy that it cant really be impacted heavily by some new data, there's not really much that could happen in the next few years that could invalidate your theory or imapct on it much. That's not because you're a baysian, it would be true even if you were a frequentists. Different theories can be far more sensitive.


I suppose it must increase with inflation but its a very small gain and the downside is you need more money in your account so the returns are smaller. Its an attractive idea is some situations, not least a high inflation world.

I am sure I was unclear. I was pointing out that we can, in real time, calculate what the prices (cost of one thing versus another) should be today and recognize disparities.

The problem is that tomorrows costs might not concur.

Necessary conditions for a variable being important for a betting strategy include the presence of it having all of the traits of varying, being measurable and having predictive power.

Liquidity doesn't have all of those traits, therefore it is just an interesting footnote. You knew just as much (for strategy's sake) without including it.

And that is what I bet on. I really don't care about my net worth next year (I have sufficient enough of a safety net built in to survive intact).

Of course, I am screwed if every company in the world goes bankrupt, but I haven't found a work-around that is affordable.

Don't agree. When you need (for future goals) for your money to grow, it is quite expensive.

The stuff I do isn't conducive to silliness. I buy stuff. Moreso when it seems cheap. Less so when it seems expensive. And when things seem very expensive, I sell a small amount and pocket the proceeds (and buy something else).

The real risk is that there might be a better strategy.

Yes, but that will come about in real time. Won't overemphasize the new data or underemphasize it.

My theory about the expected value of the market is relatively unimportant in the strategy, unless I am wrong and I am playing the wrong game.

That's all fine, not sure we disagree. its true for gold as well. The difficulties are fairly obvious, the causation force is still there just harder to exploit the more difficult it is to calculate the costs (including tomorrows costs).

Necessary conditions for a good apriori model are understanding the concepts.

It has the vital trait of being an important concept and i didn't know anything like as muich before I understood liquidity. I had the same data but data without concepts is nothing to me.

Exactly but real time is a very slow way to gather data which was my point. if we have historical data we should treat it as a precious resource as there's no easy way to replace it.