Anyone here using a RAID 0 (striped) hard disk configuration for everyday trading, eg, X_Trader, charting (but not backtesing)? Do you notice any material performance enhancement? As a general rule, does trading software access the hard disk a great deal?
Probably not. And it doubles your risk of hard disk failure (if either disk fails, you are out of action).
I'm definitely no computer expert but I'm gonna have to say no. Not a ton of hard drive activity going on for me. I personally don't think it would be worth it for trading. My computer is set up in RAID 1 configuration. Data mirroring in case one of my drives fails. Nice piece of mind as I've had a hard drive fail on me in the past.
I have RAID-0. I don't think it makes any difference. If it does it's so small you can't tell. If I had it to do over again, I would probably use RAID-1. It supposedly slows things down somewhat, but I doubt you could detect it, and it would be a good safety feature. OldTrader
If you have a fast processor and lots of RAM, probably not. But if you've ever had a HDD running swapping data or doing something right when you click on a trade, and the PC takes 5-8 seconds to finish it's previous writing before your trade is processed and sent by your PC, you've never lived Seriously, fast processor; plenty of ram first. Then do RAID 0 with 7200 or even 10K SATA drives, and you can't miss. That's what I run. Quad processor; 4 gigs of ECC ram and RAID 0 with 10K WD SATA drives. I NEVER have the PC doing something when I execute a trade.
Check your hard drive status light when you are trading, if it is constantly flashing or always on then yes will help. If it is a program running mostly in memory and that light barely moves, then no. Totally depends on how the programmers of your trading software decided to do it, some platforms swap a ton of data to the HD in which case it does help. Like has been said though, RAM, processor, and front-side bus speed will have bigger impacts so upgrade those first and see if you still need RAID before doing it. If you really want to have a CPU on steroids you can get RAID 0+1 which stripes then mirrors, giving you the RAID 0 speed along with RAID 1 reliability. Just be prepared for a cpu with 4 hard drives and that generates enough heat to make you trade naked unless you have top of the line AC (some of us trade naked anyways) .
Here is how i view the situation. In a previous chapter of life's stories i was a route salesmen delivering Coca Cola to grocery stores to feed 4 and pay for a house etc. During this same period i was taking classes on economics and was drag racing cars. Point being i worked, learned and played all at the same time to enjoy things as they were and was preparing for the changes down the road. When i raced the 1957 Chevy with a balanced, clearances, ported, polished heads, ram induction, duel quads, etc, etc, etc, Z28 engine i knew the one or two other guys in the same class were backed by sponsor's and had top flight equipment. What did that tell me to do? Very simple, go to the weekend races with a piece of equipment that was gonna make them take notice. A trading computer is basically the same deal. You use the puter for racing only, no grocery shopping (web browsing), no games etc. I viewed a puter the same as the ordinary looking "grocery getter" race car but would run the qtr mile in approx 125 mph and record times. I daytrade and will more than not use mkt orders. In an average year i will hit the mouse and send a order (LIMIT) order and miss the fill because i was a millisecond short. Yes i know i am assuming but i can assure you i would say a millisecond here, a millisecond there and i would have got a fill. Back to mkt orders, a mkt order can also depend on milliseconds. A millisecond late and you can just missed a couple hundred maybe.... The point is, GO TO THE RACE with your best machine. I know i have lost a few ticks in a years time that would have been better if the puter, the connection etc was JUST a hair better. Raid-0 is just a single example to use for a millisecond here or there. Think about going to a dance. You did not go right from work, you pampered yourself for the ladies. Were you going to the dance to dance with the wallflowers or the "Best of Breed"? Do the same with your puter. Milliseconds cound when daytrading.
I'm not a scalper with viper-like reflexes... in fact, not a scalper at all. However, my formal education included a heavy dose of logic. Neuromotor speed.... (1) If you consciously decide to execute a trade 'NOW', it takes about 30ms for the average person to execute the mouse click... such is the case for any "voluntary" neuromotor task. (2) If you have a "neuromotor reaction" to a stimulus, it takes about 75ms. An example of this would be, you're watching your stock drop and you're planning to "click off a buy as soon as you see it hit 50.75". If you clicked immediately when you saw that, it would still take 75ms to get the mouse button pushed. So, as you watch prices change real time, you can see there is lots of "noise" both up an down... pennies each way. Therefore I submit that over a hundred scalping attempts, as you're clicking off your trade, the NEXT tick or two is as likely to be in your favor as much as it is not in your favor. If that is the case, "milliseconds don't really count". Over time and a number of trades, the tick slippages should be about 50-50 and even out. No?
Point well taken. BUT!!! as a breakout trader in the ES i am looking for a breakout, a continuation etc. I might sit there and predict price will do this or that and put in a limit order to be early in the "Q" hoping for a fill. I have missed fills because the "first in first out" orders were filled before i was picked off. Ok, that is not a matter of speed, agreed. Now, if i see i missed the trade because i missed a fill on a limit order and still want in it is a split second decision if i REALLY want in still. If i indeed still want in then i have no choice to hit a Market order and accept what is given to me, like it or not. In a situation like that milliseconds matter. The slowest lapse of time in entering a market order into the mkt is still the "finger pressing the mouse" but like drag racing, i still feel better knowing i have brought my best engine to the race. Roger Penske builds a team with the best he can afford etc.
Lots of us do what you do. I suggest you examine your strategy logically.. I think you'll come to the conclusion that market orders are a better way to go all the time. A tick or two slippage in the ES also evens out over time.