Any way to profit from short-term direction predictions?

[MrBrilliant]

I'm a bit new so forgive me if this is blatantly trivial or if this turns out to be a "stupid" question, but here we go:

I wrote a computer program that can make predictions on whether or not the DOW will be up or down over the next few seconds.

At the moment the accuracy is something like 54% for 1 minute, and over 60% for 10-20 second durations.

Is there any way to profit from this?

 

[DionysusToast]

It depends on the distance it travels in that time.

Figure initially that you need to give up 1 tick on both entry and exit. Would the winners make up for the losers + commissions?

 

[MrBrilliant]

I see.. So is there any way to profit from short-term movements without having to take 1 tick both ways? (How much is a tick by the way, is it one point?)

 

[SuddenDeath]

it is impossible to predict short term price direction.

if u dont believe me check the next time there is a trading competition, 80% people end up losing.

 

[MrBrilliant]

it is impossible to predict short term price direction.

if u dont believe me check the next time there is a trading competition, 80% people end up losing.

I think you're talking about something else; by short-term I mean several seconds.

 

[MrBrilliant]

By the way, what's the most efficient (lowest spread) way to trade indices?

 

[wino59]

I see.. So is there any way to profit from short-term movements without having to take 1 tick both ways? (How much is a tick by the way, is it one point?)

Dow Futures are $5 a tick, 1 tick = 1 point - However that is a pretty thin market, so I don't know what kind of size you would be trading, the liquidity is not there.

you don't have to give up a tick in either direction if you are placing limit orders and not market orders. (but that doesn't mean you will get filled either) especially in thinner markets.

So basically you need to be at least making 1 point to cover your commissions, and and I am not sure what you are using for a loss, and risk reward scenario, so it is hard to tell if would be profitable.

 

[MrBrilliant]

When you say there is no guarantee that my order will be filled, does that mean it would be possible to open a position, and then have trouble closing it?

 

[wino59]

When you say there is no guarantee that my order will be filled, does that mean it would be possible to open a position, and then have trouble closing it?

No. you can always place a market order to close. Limit orders are the ones shown on the DOM the bids and offers, so if you were trading the DOW futures then it would be something like this.

Let say the price is 14500

Bid is 14500 and the offer is 14501 you could buy 14501 right now, and get your fill provided you are not buying more contracts than what is offered

you can buy 14500 if you place a bid, but you need people to sell into you to get filled, there may be 200 bids, and only 100 sell into the bid, then you may not get filled if you were 101 or higher in line.

I would say the only time you would have a hard time closing a position is if you were trading more size then the liquidity. Meaning wanting to trade more contracts than what is offered.

 

[Splitlink]

I think you're talking about something else; by short-term I mean several seconds.

I must be getting old. I don't see how anyone can predict that, frankly. Even, just thinking about it takes some seconds, at least. By that time the game is over. Asking anyone for advice on how to do it is like looking for a needle in a haystack.

It has to be computerised trading and is, completely, random. In addition, you are guessing that the market won't spike at rhe same time as your programmed trade opens,

 

[lurkerlurker]

it is impossible to predict short term price direction.

if u dont believe me check the next time there is a trading competition, 80% people end up losing.

I don't see how anyone can predict that, frankly. Even, just thinking about it takes some seconds, at least.

Asking anyone for advice on how to do it is like looking for a needle in a haystack.

It has to be computerised trading and is, completely, random.

Yes, you can trade for holding times less than a minute and be very profitable, both manually and with algos.

I have a trading acquaintance who makes several million dollars per year with one of his strategies - an automation which scalps volatile markets for ticks. Risk is minimal and the trades are opened and closed in 1-5 second bursts throughout the day, for 1 to a few ticks profit, in sizes of 10-100 futures contracts. His model has extremely good predictive power for the next few seconds. I know his bona fides: ex institutional at an IB back east, founded his own B/D in Chi, runs his own trading team, and owns several exchange memberships.

Few here know anything about doing this because successful traders have very little incentive to contribute. Those who have tried to do so have been discouraged. Therefore it is impossible to get serious answers to relevant questions, which makes this site worse than useless for aspiring traders. So unfortunately he isn't here to answer your questions. And it appears neither is anyone else who specialises in this niche. Its apparent that any industry practitioner is an unwanted pollutant on financial discussion boards - the above quoted sums up the attitude and quality of contribution on sites such as trade2win.

When I needed a new clearing firm, this acquaintance took my call, listened to my requirements, explained a few little known things about the exchange membership and clearing process which were helpful, and offered to introduce me to a C-level exec at a major clearing firm which would be able to meet my requirements. Why? Because he is helpful and I was polite, sincere, respectful, genuine, diligent...and most importantly didn't act like a complete ass.

There is lots of information available, and if you're decent you might be able to get an overview from one of the many folks in this industry who are generous enough to give a sincere aspirant a helping hand. If you can get an introduction. And that is the worst part about the negative attitudes here: everybody loses as the serious professionals have every disincentive to participate, even when they are naturally predisposed to helping others. Building relationships is key in this as well as other businesses. At a certain level,you will need your clearing firm, exchange and perhaps even regulator to look favourably on what you are doing.

We have this great technology...an opportunity, unprecedented, to publish information and help others through sharing knowledge...and we turn it into an entertainment device, and a sandpit to vent the worst parts of our character...greed, jealousy, closed mindedness, anger...I've used it to connect with experts who could teach me what I wanted to learn...you guys have piss3ed on the experts and therefore poisoned the well for those who come later.

When you abuse contributors, or provide closed mindedness, jealousy, and negative vibes, you alienate yourselves from anyone who is or could be successful. This behaviour isn't logical, so I presume it is a latent tendency brought out by the anonymous nature of Internet discussion sites.

Something to think about as DaxDayTrader30 joins the legion of intelligent, thoughtful, serious contributors to quit trade2win due to the prevailing standards of behavior. The fact that these points aren't required reading for all members, along with a strict code of conduct enforced by immediate bans, is all the evidence I and others need to see that this site is a product for advertisers, not a trading community. The policy of the owner drives traffic and generates revenue, but makes a sterile and useless environment for traders or aspiring traders.

 

[scholfield]

Something to think about as DaxDayTrader30 joins the legion of intelligent, thoughtful, serious contributors to quit trade2win due to the prevailing standards of behavior. The fact that these points aren't required reading for all members, along with a strict code of conduct enforced by immediate bans, is all the evidence I and others need to see that this site is a product for advertisers, not a trading community. The policy of the owner drives traffic and generates revenue, but makes a sterile and useless environment for traders or aspiring traders.

Turns out Daxdaytrader30 was infact none other than king of the scammers, Mrspreadbet!

 

[bigdady]

When you say there is no guarantee that my order will be filled, does that mean it would be possible to open a position, and then have trouble closing it?

Yes It is impossible to say exactly when you be filled (other trader willing to pay the offer price, or the higher price ) on a limit order. If exactly at that moment, the bid/offer move down 1 tic as you decide you want to sell, and you don't adjust your limit price, then basically, the market might be running from you. What you have to do, if you want open and close a position rapidly (depending how program it), you have to take into consideration the possibly you don't have a buyer, and you sell at a loss, hit your stop basically.

 

[MrBrilliant]

Thanks lurkerlurker, I appreciate your post.

Also, thanks bigdady and wino for the insight into how ticks and orders work.

As I mentioned, my program seems to be predicting the direction correctly about 60% of the time for 10-20 second durations, so I disagree with the idea that it's impossible to make predictions. It may be the case that I won't be able to make any money from the strategy, due to commissions/spreads/liquidity issues eating up the profits... and that's what I'm trying to figure out.

I don't know very much about the details of trading (so far I've just done spread-betting), so I don't know how to actually implement the strategy in a real market.

Can anyone give me a nudge in the right direction: What's the lowest-spread way to test out an automated strategy with low contract sizes?

Thanks guys

 

[bigdady]

Thanks lurkerlurker, I appreciate your post.

Also, thanks bigdady and wino for the insight into how ticks and orders work.

As I mentioned, my program seems to be predicting the direction correctly about 60% of the time for 10-20 second durations, so I disagree with the idea that it's impossible to make predictions. It may be the case that I won't be able to make any money from the strategy, due to commissions/spreads/liquidity issues eating up the profits... and that's what I'm trying to figure out.

I don't know very much about the details of trading (so far I've just done spread-betting), so I don't know how to actually implement the strategy in a real market.

Can anyone give me a nudge in the right direction: What's the lowest-spread way to test out an automated strategy with low contract sizes?

Thanks guys

I would say try SPY on NYSENYSE would lowest spread ans cost.

 

[DionysusToast]

Thanks lurkerlurker, I appreciate your post.

Also, thanks bigdady and wino for the insight into how ticks and orders work.

As I mentioned, my program seems to be predicting the direction correctly about 60% of the time for 10-20 second durations, so I disagree with the idea that it's impossible to make predictions. It may be the case that I won't be able to make any money from the strategy, due to commissions/spreads/liquidity issues eating up the profits... and that's what I'm trying to figure out.

I don't know very much about the details of trading (so far I've just done spread-betting), so I don't know how to actually implement the strategy in a real market.

Can anyone give me a nudge in the right direction: What's the lowest-spread way to test out an automated strategy with low contract sizes?

Thanks guys

A bit hard to say when you didn't mention which sorts of markets you are looking at - stocks, futures, forex.

If stocks indices as implied by the forum you are in, then you need to figure out if you can enter at limit or not because then you get paid to trade.

The limit/market thing is key to your strategy working as is the actual size of the move you are predicting.

You would really need to put some numbers up in order to get an intelligent response.

 

[MrBrilliant]

A bit hard to say when you didn't mention which sorts of markets you are looking at - stocks, futures, forex.

If stocks indices as implied by the forum you are in, then you need to figure out if you can enter at limit or not because then you get paid to trade.

The limit/market thing is key to your strategy working as is the actual size of the move you are predicting.

You would really need to put some numbers up in order to get an intelligent response.

Thanks for your response.

Yes, it's indices, Dow Jones at the moment but I could look at other things.

Sure, here are some numbers: I've tested my strategy on the DJI tick-data; most of the predictions are between 2 and 10 points over 10-20 seconds. So for example, I get a "buy" signal on the DJI when it's at 15,200 and then 15 seconds later it's at 15,205.

I don't know how that translates to ticks or pips?

 

[hb7of9]

it is impossible to predict short term price direction.

if u dont believe me check the next time there is a trading competition, 80% people end up losing.

interesting POV, I also found this opposite POV interesting.

I also found this one interesting, from the New Market Wizards by Jack Schwager.

"My niche is short-term trading, which is how I make my bread and butter. The occasional long-term trades
are frosting on the cake. I believe that only short-term price swings can be predicted with any precision. The
accuracy of a prediction drops off dramatically, the more distant the forecast time. I'm a strong believer in
chaos theory.
[A basic concept of chaos theory is that for aperiodic systems-i.e., systems that never exactly repeat
themselves and hence never find a steady state, such as weather or the markets-slight differences in variable
values or measurements can be magnified to have huge effects over increasing., periods of time. The
technical name for this phenomenon-sensitive dependence on initial conditions-has become better known as
the Butterfly Effect"

 

[DionysusToast]

Thanks for your response.

Yes, it's indices, Dow Jones at the moment but I could look at other things.

Sure, here are some numbers: I've tested my strategy on the DJI tick-data; most of the predictions are between 2 and 10 points over 10-20 seconds. So for example, I get a "buy" signal on the DJI when it's at 15,200 and then 15 seconds later it's at 15,205.

I don't know how that translates to ticks or pips?

Points as you have described = ticks on the DOW. Pips is a term generally reserved for forex.

You may have something workable but the DOW Futures is pretty thin and so you could be facing slippage on that (although that is by no means a certainty as we don't know what you are doing). On the bright side you'd more than likely get a fill at your target.

Also with DOW futures, the commissions are pretty high relative to the value of a tick.

So - a thicker market might be better but what you do might not work there because of the different liquidity picture. Some markets only cost $2 per round turn and pay off $32 a tick. DOW is probably going to cost you $4 per round turn and pay off $5 a tick.

DOW is also not very scalable but you might want to look at your methods on crude as it is think and whippy but you'd be able to trade more size. Of course - on ES and US Treasuries, you could swing a lot of size.

Some other stats that you need to look at:
- average move against in a winning trade (MAE - maximum adverse excursion)
- some estimation of the size of your losers

 

[Liquid validity]

All depends on how its been tested,
If it was tested on native tick data (data on own PC),
then connection latency and fill speed in real market conditions
haven't been accounted for.
In other words, testing may well be flawed.
If it hasn't even been tested on bid / ask tick data, but mid m1 or something,
again massively flawed due to high value timeseries data with no spread.

Even if the test is SIM and has been conducted on live data,
fill speed, missed fills (limits) / slippage (market) haven't been accounted for.
Missed fills and slippage will have a massive impact with a very short trade duration regardless.

With this kind of thing, only an outright live test has any worth.
Using a retail net connection and other retail bottlenecks,
I'd be surprised if you still come out ahead of the latency arb co-location
HFT outfits.
Like I say, all depends on the answers to the above questions.

With this kind of stuff, connection latency, fill speed,
software calculation and execution speed and retail comms are the main hurdles,
and can't be ignored, they are just as important, if not more so than the core code.

At the very least, to have half a chance, you'd need X-trader,
low latency connection (ideally co-location) and commission volume discount.
All of which by definition adds to your costs and thus increases your breakeven size,
which in turn bumps up the bankroll requirements massively.
Even then, you are still near the back of the queue in terms of costs and technology.

Its perfectly possible to have something that appears profitable,
but without the above points being part of the test,
will most likely fail under real market conditions.