https://github.com/holgern/steemrewarding
steemrewarding.com is a feature-rich automatic voting tool. It can be used to create voting rules at https://steemrewarding.com, when posting authority was given to the 
steemrewarding is currently used by 126 users which created 1193 rules for posts, 85 rules for comments and 24 trail vote rules. In the last 7 days were 4244 time based votes and 1986 vp based votes broadcasted through steemrewarding.
When entering a steem account name after stemrewarding.com, a new rule can be directly be created. For example, https://steemrewarding.com/@utopian-io
will open the new rule editor:
A user of steemrewarding.com can now optimize the vote delay time for all time based vote rules automatically. The optimization must be enabled in https://steemrewarding.com/settings:
When enabled, each vote_log is analyed in the calc_curation_performance.py script.
best_performance = 0
best_vote_delay_min = 0
for v in c["active_votes"]:
v_SBD = stm.rshares_to_sbd(int(v["rshares"]))
if v_SBD > 0 and int(v["rshares"]) > rshares * 0.5:
p = float(curation_rewards_SBD["active_votes"][v["voter"]]) / v_SBD * 100
if p > best_performance:
best_performance = p
best_vote_delay_min = ((v["time"]) - c["created"]).total_seconds() / 60
All given votes to the upvoted post/comment that have at least half the rshares of the rule creator are analyed. When there is a single vote that have a better performance, its vote delay time and its performance is stored.
The values are shown in the vote log https://steemrewarding.com/show_vote_log:
A vote delay time is optimized when
minimum_vote_delay and maximum_vote_delay can be set at steemrewarding.com/settings
When a best_performance and best_vote_delay_min was found for a vote rule log entry, the vote delay is optimized by:
minimum_vote_delay = acc_data["minimum_vote_delay"]
maximum_vote_delay = acc_data["maximum_vote_delay"]
optimize_threshold = 1 + (acc_data["optimize_threshold"] / 100)
optimize_ma_length = acc_data["optimize_ma_length"]
vote_rule = voteRulesTrx.get(vote_log["voter"], c["author"], c.is_main_post())
if vote_rule is not None and not vote_rule["disable_optimization"]:
vote_delay_min = vote_rule["vote_delay_min"]
if best_performance > performance * optimize_threshold and vote_delay_min <= maximum_vote_delay and vote_delay_min >= minimum_vote_delay:
if optimize_ma_length > 1:
vote_delay_min = (vote_delay_min * (optimize_ma_length - 1) + best_vote_delay_min) / optimize_ma_length
else:
vote_delay_min = best_vote_delay_min
if vote_delay_min > maximum_vote_delay:
vote_delay_min = maximum_vote_delay
elif vote_delay_min < minimum_vote_delay:
vote_delay_min = minimum_vote_delay
voteRulesTrx.update({"voter": vote_log["voter"], "author": c["author"], "main_post": c.is_main_post(),
"vote_delay_min": vote_delay_min})
vote_log["optimized_vote_delay_min"] = vote_delay_min
When the optimal vote delay differs less than optimize_threshold percentage from the old one, the vote delay is not changed. This prevents that the vote delay time from jittering.
A simplified moving average calculation is then used to update the vote delay time. The influence of the optimal voting time is set by optimize_ma_length. When it is set to 20, it will take 20 steps approach a new optimum.
He is an example, that show how it works:
I set the vote delay to 13 minutes, the best vote delay is at 8.2 minutes. The new vote delay is then:
(( optimize_ma_length - 1) * 13 min + 8.2 min) / optimize_ma_length
which is for optimize_ma_length=20 12.76 minutes. This means that the vote delay for my rule is reduced in order to approach the optimal vote time.
The parameter optimize_ma_length and optimize_threshold can be set in steemrewarding.com/settings