Announcement

Collapse

Announcement

~ gang gang ~
See more
See less

just some jim simons flavored food for thought on risk / reward

Collapse
X
Collapse
 
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
Previous template Next
  • #1

    just some jim simons flavored food for thought on risk / reward

    his fund is averaging 66% returns every year but loses on 49% of their trades.

    they cut losses fast when their model doesnt shake out, and when it does shake out they bink a few points. and it adds up.



    Code:
    roll = 10000

print("Starting roll: ", roll)

start = roll

# place 98 bets. 49 lose 1%, 49 gain 2%.
for n in range(49):
# we enter a trade and lose 1%
roll = (roll * .99)
# we enter a trade and win 2%
roll = (roll * 1.02)



# bets 50 & 51, both gain 2%
roll = (roll * 1.02)

roll = (roll * 1.02)


print("Final roll: ", roll)

print("Gains: ", ( roll / start ))
    $ python3 ./simons.py

    Starting roll: 10000

    Final roll: 16777.733632129108

    Gains: 1.6777733632129108



    easy mark, sfo.
    Tags: None
  • #2
    That script follows a deterministic pattern where you're always losing 1% on a trade and then gaining 2% on the next trade, in that order.

    Let's leave it to chance:

    Code:
    import random

bankroll = 10000

print("Starting roll: ", bankroll)

start = bankroll

# place 98 bets. 49 lose 1%, 49 gain 2%.
for n in range(98):

diceroll = random.randint(1, 6)

if diceroll <= 3:
# we enter a trade and lose 1%
bankroll = (bankroll * .99)
else:
# we enter a trade and win 2%
bankroll = (bankroll * 1.02)


print("Final roll: ", bankroll)

print("Gains: ", ( bankroll / start ))
    Code:
    Starting roll: 10000
Final roll: 12700.275275779037
Gains: 1.2700275275779036
    That's 98 trades with an equal chance of being either a 1% loss or a 2% gain. A few more runs...

    Code:
    Starting roll: 10000
Final roll: 15191.581828878534
Gains: 1.5191581828878533
    Code:
    Starting roll: 10000
Final roll: 13890.186143507228
Gains: 1.3890186143507228
    Code:
    Starting roll: 10000
Final roll: 12326.737767667879
Gains: 1.232673776766788

    Comment

    • #3
      Okay now let's run that 50,000 times.

      Code:
      import random
import statistics

start = 10000

print("Starting roll: ", start)

results = []

def tradesimulation():

bankroll = start

# place 98 bets. 49 lose 1%, 49 gain 2%.
for n in range(98):
diceroll = random.randint(1, 6)

if diceroll <= 3:
# we enter a trade and lose 1%
bankroll = (bankroll * .99)
else:
# we enter a trade and win 2%
bankroll = (bankroll * 1.02)

bankroll = round(bankroll, 1)

results.append(bankroll)


for i in range(50000):
tradesimulation()

print(results)
print("\n")
print(f"Median: {statistics.median(results)}")
      Click image for larger version Name: Annotation 2020-08-09 174154.png Views: 0 Size: 89.2 KB ID: 2153

      I don't know the mathematical way of calculating the average gain after 98 such trades, but I think this brute force approach gets within 1 decimal place of the correct answer.

      Comment

      • #4
        id say the real lesson here shouldnt be 'it is or isnt a -1% / +2% binary formulae', its simply to demonstrate how lucrative it is to embrace conservative risk profiles. to make 66% (or 61.26%) profits a year while losing on 49% of your positions is powerful.

        Comment

        • #5
          Isn't that just what HFT is about?

          edit: also what built Vegas

          Comment

          • #6
            i mean, HFT can mean a lot of things at this point. on the one hand its citadel performing microsecond fast arbs, on the other hand its citadel performing microsecond fast front running attacks on inbound orders buy purchasing all the stock available in their dark pool and then kicking up the ask on them by a cent or two.

            i was told that i was a HFT because i wanted more insight into how to feed minute by minute ticks into my alg, hilariously.

            in any case its tempting to oversimplify simons point here but again this needs a spotlight on it; his fund outperforms literally every other fund on earth, and does so while scaling up in a way that would break most other algs. shits pretty impressive. like you generally dont run up big losses if you dont take big losses. thats kinda revelatory.
            • Likes 1

            Comment

            • #7
              Yea saying a variant of HFT would've been more precise. One of the non scummy ways. Isn't purely just taxing retail.
              • Likes 1

              Comment

              • #8
                Guessing you can't do the same as retail though.

                Comment

                • #9
                  https://youtu.be/jWsx2iqO1ks
                  • Likes 1

                  Comment

                  • #10
                    Originally posted by gimmick View Post
                    Guessing you can't do the same as retail though.

                    i mean, my end game is essentially HFT; i want to automate entry and exit on high probability scalps, ultimately on e-minis. so the notion of value / risk aversion is fairly nonstandard. jim simon's off hand comment about the 49/51 win ratio is a bit revelatory for that reason, particularly in the context of futures, because after you factor in slippage and commissions, the binary nature of scalping e-minis plus the unbelievable margin leverage available produces some absolutely shocking profit models.

                    the industrial HFT sector essentially operates on the same principle but with exponentially more trades in a micro market they essentially own.

                    but i think there is room for adoption in the retail ecosystem if one scales up volume, scales back exposure, and keys in on established market trends via indicator fu.
                    • Likes 1

                    Comment

                    • #11
                      regarding hft this is really great:

                      https://www.youtube.com/watch?v=aq1Ln1UCoEU

                      not a ton of new information but im reading the biography of nav sarao so its insanely interesting stuff for me atm.

                      the opinions that no one controls the market essentially have perhaps aged poorly.
                      • Likes 1

                      Comment

                      Previous template Next
                      Working...
                      X