wqb
/
alpha_tools
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

++

Browse Source
main
Jack 4 weeks ago
parent 0ef8731055
commit 88ac366352
2 changed files with 155 additions and 49 deletions
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Show Stats Download Patch File Download Diff File
  1. 9
      alpha_check/main.go
  2. 165
      alpha_check/main.py

9
alpha_check/main.go
Unescape View File

@ -6,7 +6,6 @@ import (
"fmt" "fmt"
"math" "math"
"strconv" "strconv"
"strings"
"time" "time"
"github.com/valyala/fasthttp" "github.com/valyala/fasthttp"
@ -59,7 +58,11 @@ type PnlResponse struct {
func NewClient(username, password string) *Client { func NewClient(username, password string) *Client {
return &Client{ return &Client{
client: &fasthttp.Client{}, client: &fasthttp.Client{
ReadTimeout: 60 * time.Second,
WriteTimeout: 10 * time.Second,
MaxIdleConnDuration: 120 * time.Second,
},
username: username, username: username,
password: password, password: password,
} }
@ -205,7 +208,7 @@ func GetAlphas(c *Client, startDate, endDate string, sharpeTh, fitnessTh float64
for _, item := range alphaResp.Results { for _, item := range alphaResp.Results {
alphaID := item.ID alphaID := item.ID
name := item.Name _ = item.Name // avoid unused variable
dateCreated := item.DateCreated dateCreated := item.DateCreated
sharpe := getFloat(item.Is, "sharpe") sharpe := getFloat(item.Is, "sharpe")
fitness := getFloat(item.Is, "fitness") fitness := getFloat(item.Is, "fitness")

165
alpha_check/main.py
Unescape View File

@ -17,50 +17,89 @@ def login():
limits=limits, limits=limits,
transport=transport transport=transport
) )
print("[INFO] 正在登录...")
response = s.post('https://api.worldquantbrain.com/authentication') response = s.post('https://api.worldquantbrain.com/authentication')
print(response.content) print(f"[INFO] 登录响应: {response.content}")
return s return s
def wait_get(s, url: str, max_retries: int = 10): def wait_get(s, url: str, max_retries: int = 10):
print(f"[WAIT_GET] 开始请求: {url[:80]}...")
retries = 0 retries = 0
while retries < max_retries: while retries < max_retries:
while True: print(f"[WAIT_GET] 第 {retries + 1} 次尝试...")
simulation_progress = s.get(url) simulation_progress = s.get(url)
if simulation_progress.headers.get("Retry-After", 0) == 0: status_code = simulation_progress.status_code
break content_length = simulation_progress.headers.get("content-length", "unknown")
time.sleep(float(simulation_progress.headers["Retry-After"])) retry_after = simulation_progress.headers.get("Retry-After", 0)
if simulation_progress.status_code < 400: print(f"[WAIT_GET] 状态码: {status_code}, content-length: {content_length}, retry-after: {retry_after}")
# 检查 Retry-After 头（即使状态码是200也可能需要等待）
if retry_after and float(retry_after) > 0:
wait_time = float(retry_after)
print(f"[WAIT_GET] 需要等待 {wait_time} 秒后重试...")
time.sleep(wait_time)
continue # 不增加重试计数，直接再试
# 成功且有内容
if status_code < 400:
# 检查 content-length
if content_length and content_length != "0":
print(f"[WAIT_GET] 请求成功，有数据")
break break
else: else:
time.sleep(2 ** retries) # 200 但无内容，可能是数据还没准备好，指数退避等待
wait_time = 2 ** retries
print(f"[WAIT_GET] 状态码200但无内容，等待 {wait_time} 秒后重试...")
time.sleep(wait_time)
retries += 1
continue
# 404 - 资源不存在，不需要重试
if status_code == 404:
print(f"[WAIT_GET] 404 Not Found，跳过重试")
break
# 429 - Rate Limit
if status_code == 429:
print(f"[WAIT_GET] 429 Rate Limit")
continue
# 其他错误，指数退避重试
wait_time = 2 ** retries
print(f"[WAIT_GET] 请求失败，等待 {wait_time} 秒后重试...")
time.sleep(wait_time)
retries += 1 retries += 1
return simulation_progress return simulation_progress
def get_alphas(start_date, end_date, sharpe_th, fitness_th, region, alpha_num, usage): def get_alphas(s, start_date, end_date, sharpe_th, fitness_th, region, alpha_num, usage):
s = login() print(f"[INFO] 开始获取Alpha列表, 参数: start_date={start_date}, end_date={end_date}, sharpe_th={sharpe_th}, fitness_th={fitness_th}, region={region}, alpha_num={alpha_num}, usage={usage}")
output = [] output = []
count = 0 count = 0
for i in range(0, alpha_num, 100): for i in range(0, alpha_num, 100):
print(i) print(f"[GET_ALPHAS] 处理偏移量: {i}")
url_e = "https://api.worldquantbrain.com/users/self/alphas?limit=100&offset=%d" % (i) \ url_e = "https://api.worldquantbrain.com/users/self/alphas?limit=100&offset=%d" % (i) \
+ "&status=UNSUBMITTED%1FIS_FAIL&dateCreated%3E=2025-" + start_date \ + "&status=UNSUBMITTED%1FIS_FAIL&dateCreated%3E=" + start_date \
+ "T00:00:00-04:00&dateCreated%3C2025-" + end_date \ + "T00:00:00-04:00&dateCreated%3C=" + end_date \
+ "T00:00:00-04:00&is.fitness%3E" + str(fitness_th) + "&is.sharpe%3E" \ + "T00:00:00-04:00&is.fitness%3E" + str(fitness_th) + "&is.sharpe%3E" \
+ str(sharpe_th) + "&settings.region=" + region + "&order=-is.sharpe&hidden=false&type!=SUPER" + str(sharpe_th) + "&settings.region=" + region + "&order=-is.sharpe&hidden=false&type!=SUPER"
url_c = "https://api.worldquantbrain.com/users/self/alphas?limit=100&offset=%d" % (i) \ url_c = "https://api.worldquantbrain.com/users/self/alphas?limit=100&offset=%d" % (i) \
+ "&status=UNSUBMITTED%1FIS_FAIL&dateCreated%3E=2025-" + start_date \ + "&status=UNSUBMITTED%1FIS_FAIL&dateCreated%3E=" + start_date \
+ "T00:00:00-04:00&dateCreated%3C2025-" + end_date \ + "T00:00:00-04:00&dateCreated%3C=" + end_date \
+ "T00:00:00-04:00&is.fitness%3C-" + str(fitness_th) + "&is.sharpe%3C-" \ + "T00:00:00-04:00&is.fitness%3C-" + str(fitness_th) + "&is.sharpe%3C-" \
+ str(sharpe_th) + "&settings.region=" + region + "&order=is.sharpe&hidden=false&type!=SUPER" + str(sharpe_th) + "&settings.region=" + region + "&order=is.sharpe&hidden=false&type!=SUPER"
urls = [url_e] urls = [url_e]
if usage != "submit": if usage != "submit":
urls.append(url_c) urls.append(url_c)
for url in urls: for url in urls:
print(f"[GET_ALPHAS] 请求URL: {url[:100]}...")
response = s.get(url) response = s.get(url)
print(f"[GET_ALPHAS] 响应状态码: {response.status_code}")
try: try:
alpha_list = response.json()["results"] alpha_list = response.json()["results"]
print(f"[GET_ALPHAS] 获取到 {len(alpha_list)} 个alpha")
for j in range(len(alpha_list)): for j in range(len(alpha_list)):
alpha_id = alpha_list[j]["id"] alpha_id = alpha_list[j]["id"]
name = alpha_list[j]["name"] name = alpha_list[j]["name"]
@ -78,7 +117,7 @@ def get_alphas(start_date, end_date, sharpe_th, fitness_th, region, alpha_num, u
if sharpe < -sharpe_th: if sharpe < -sharpe_th:
exp = "-%s" % exp exp = "-%s" % exp
rec = [alpha_id, exp, sharpe, turnover, fitness, margin, dateCreated, decay] rec = [alpha_id, exp, sharpe, turnover, fitness, margin, dateCreated, decay]
print(rec) print(f"[GET_ALPHAS] 符合条件的alpha: {rec}")
if turnover > 0.7: if turnover > 0.7:
rec.append(decay * 4) rec.append(decay * 4)
elif turnover > 0.6: elif turnover > 0.6:
@ -92,16 +131,18 @@ def get_alphas(start_date, end_date, sharpe_th, fitness_th, region, alpha_num, u
elif turnover > 0.3: elif turnover > 0.3:
rec.append(decay + 2) rec.append(decay + 2)
output.append(rec) output.append(rec)
except: except Exception as e:
print("%d finished re-login" % i) print(f"[GET_ALPHAS] 处理第 {i} 个时出错: {e}, 重新登录...")
s = login() s = login()
print("count: %d" % count) print(f"[GET_ALPHAS] 总共获取 {count} 个alpha, 符合条件的有 {len(output)} 个")
return output, s return output, s
def check_consecutive_non_zero_values(alpha_id, data, required_streak=200): def check_consecutive_non_zero_values(alpha_id, data, required_streak=200):
print(f"[CHECK_NON_ZERO] 检查alpha {alpha_id}, 数据长度: {len(data) if data else 0}")
if not data or len(data) < required_streak: if not data or len(data) < required_streak:
print(f"[CHECK_NON_ZERO] 数据不足{required_streak}条, 跳过检查")
return True return True
def check_column(column_data): def check_column(column_data):
@ -133,30 +174,65 @@ def check_consecutive_non_zero_values(alpha_id, data, required_streak=200):
column1_values.append(row[1]) column1_values.append(row[1])
column2_values.append(row[2]) column2_values.append(row[2])
print(f"[CHECK_NON_ZERO] 列1数据点数: {len(column1_values)}, 列2数据点数: {len(column2_values)}")
if column1_values and column2_values: if column1_values and column2_values:
is_col1_all_zeros = all(v == 0 for v in column1_values) is_col1_all_zeros = all(v == 0 for v in column1_values)
is_col2_all_zeros = all(v == 0 for v in column2_values) is_col2_all_zeros = all(v == 0 for v in column2_values)
if is_col1_all_zeros or is_col2_all_zeros: if is_col1_all_zeros or is_col2_all_zeros:
print(alpha_id, "不合法") print(f"[CHECK_NON_ZERO] {alpha_id} 不合法: 存在全零列")
return False return False
if not check_column(column1_values): if not check_column(column1_values):
print(alpha_id, "不合法") print(f"[CHECK_NON_ZERO] {alpha_id} 不合法: 列1存在连续非零值")
return False return False
if not check_column(column2_values): if not check_column(column2_values):
print(alpha_id, "不合法") print(f"[CHECK_NON_ZERO] {alpha_id} 不合法: 列2存在连续非零值")
return False return False
print(f"[CHECK_NON_ZERO] {alpha_id} 通过检查")
return True return True
def get_alpha_pnl_legal(s, alpha_id: str) -> bool: def get_alpha_pnl_legal(s, alpha_id: str) -> bool:
print(f"\n[PNL_CHECK] ===== 开始检查Alpha: {alpha_id} =====")
not_legal_id = [] not_legal_id = []
pnl = wait_get(s, "https://api.worldquantbrain.com/alphas/" + alpha_id + "/recordsets/pnl").json() print(f"[PNL_CHECK] 正在获取PNL数据...")
records = pnl["records"] resp = wait_get(s, "https://api.worldquantbrain.com/alphas/" + alpha_id + "/recordsets/pnl")
# 检查状态码
if resp.status_code == 404:
print(f"[PNL_CHECK] {alpha_id} PNL接口返回404，可能Alpha不存在或无权限，判定为不合法")
return False
if resp.status_code >= 400:
print(f"[PNL_CHECK] {alpha_id} PNL接口返回错误状态码 {resp.status_code}，判定为不合法")
return False
# 检查响应内容是否为空
if not resp.content or len(resp.content) == 0:
print(f"[PNL_CHECK] 警告: {alpha_id} 响应内容为空！")
print(f"[PNL_CHECK] 状态码: {resp.status_code}")
print(f"[PNL_CHECK] 响应头: {dict(resp.headers)}")
print(f"[PNL_CHECK] 此Alpha可能是好的，但无法获取PNL数据，跳过处理（不隐藏）")
return True # 返回True表示不处理，不隐藏
try:
pnl = resp.json()
except Exception as e:
print(f"[PNL_CHECK] 警告: {alpha_id} JSON解析失败: {e}")
print(f"[PNL_CHECK] 响应内容前500字符: {resp.text[:500]}")
print(f"[PNL_CHECK] 状态码: {resp.status_code}")
print(f"[PNL_CHECK] 响应头: {dict(resp.headers)}")
print(f"[PNL_CHECK] 此Alpha可能是好的，但无法解析PNL数据，跳过处理（不隐藏）")
return True # 返回True表示不处理，不隐藏
records = pnl.get("records", [])
print(f"[PNL_CHECK] 获取到 {len(records)} 条PNL记录")
if not records: if not records:
print(f"[PNL_CHECK] {alpha_id} 无PNL记录, 判定为不合法")
return False return False
date_list = [] date_list = []
@ -165,16 +241,21 @@ def get_alpha_pnl_legal(s, alpha_id: str) -> bool:
date_obj = datetime.datetime.strptime(record[0], '%Y-%m-%d').date() date_obj = datetime.datetime.strptime(record[0], '%Y-%m-%d').date()
date_list.append(date_obj) date_list.append(date_obj)
except Exception: except Exception:
print(f"[PNL_CHECK] {alpha_id} 日期解析失败, 判定为不合法")
return False return False
min_date = min(date_list) min_date = min(date_list)
max_date = max(date_list) max_date = max(date_list)
total_days = (max_date - min_date).days total_days = (max_date - min_date).days
print(f"[PNL_CHECK] 日期范围: {min_date} 到 {max_date}, 总天数: {total_days}")
if total_days < 2920: if total_days < 2920:
print(f"[PNL_CHECK] {alpha_id} 时间跨度不足8年({total_days}天), 判定为不合法")
return False return False
zero_streak_threshold = 5 * 252 zero_streak_threshold = 5 * 252
col1_zeros = [record[1] == 0 for record in records] col1_zeros = [record[1] == 0 for record in records]
print(f"[PNL_CHECK] 列1零值数量: {sum(col1_zeros)}/{len(col1_zeros)}")
def max_consecutive_zeros(arr): def max_consecutive_zeros(arr):
max_streak = current_streak = 0 max_streak = current_streak = 0
@ -184,39 +265,61 @@ def get_alpha_pnl_legal(s, alpha_id: str) -> bool:
return max_streak return max_streak
col1_max_zero_streak = max_consecutive_zeros(col1_zeros) col1_max_zero_streak = max_consecutive_zeros(col1_zeros)
print(f"[PNL_CHECK] 列1最大连续零值: {col1_max_zero_streak}")
if col1_max_zero_streak >= zero_streak_threshold: if col1_max_zero_streak >= zero_streak_threshold:
print(f"{alpha_id} 不合法：存在连续{zero_streak_threshold // 252}年零值") print(f"[PNL_CHECK] {alpha_id} 不合法：存在连续{zero_streak_threshold // 252}年零值")
not_legal_id.append(str(alpha_id)) not_legal_id.append(str(alpha_id))
return False return False
print(f"[PNL_CHECK] 正在检查连续非零值...")
if not check_consecutive_non_zero_values(alpha_id, records): if not check_consecutive_non_zero_values(alpha_id, records):
return False return False
print(f"[PNL_CHECK] {alpha_id} 合法")
return True return True
def mute(s, alpha_id): def mute(s, alpha_id):
print(f"[MUTE] 正在隐藏Alpha: {alpha_id}")
url = "https://api.worldquantbrain.com/alphas/" + alpha_id url = "https://api.worldquantbrain.com/alphas/" + alpha_id
data = { data = {
"hidden": True "hidden": True
} }
response = s.patch(url, json=data) response = s.patch(url, json=data)
print(f"[MUTE] 隐藏 {alpha_id} 完成, 状态码: {response.status_code}")
def main(): def main():
fo_tracker, s = get_alphas('12-01', '12-31', 1, 0.5, 'USA', 1000, 'submit') print("=" * 60)
print("[MAIN] 程序启动")
print("=" * 60)
s = login()
fo_tracker, _ = get_alphas(s, '2025-09-17', '2026-12-31', 1, 0.5, 'USA', 1000, 'submit')
f_num = len(fo_tracker) f_num = len(fo_tracker)
print(f_num, "个alpha 进行pnl合法检测，请耐心等待") print(f"\n[MAIN] 共 {f_num} 个alpha 进行pnl合法检测，请耐心等待")
print(f"[MAIN] 实际列表长度: {len(fo_tracker)}")
count = 0 count = 0
print(len(fo_tracker)) total = len(fo_tracker)
for i in fo_tracker[::-1][0:]: for i in fo_tracker[::-1][0:]:
if count % 25 == 0: if count % 25 == 0:
print('===========', count, '===========') print(f'\n[MAIN] =========== 进度: {count}/{total} ({count/total*100:.1f}%) ===========')
count += 1 count += 1
if get_alpha_pnl_legal(s, i[0]) == False:
print(i[0], '已经隐藏') alpha_id = i[0]
mute(s, i[0]) print(f"\n[MAIN] 正在处理第 {count} 个alpha: {alpha_id}")
if get_alpha_pnl_legal(s, alpha_id) == False:
print(f"[MAIN] {alpha_id} 检测结果: 不合法, 执行隐藏操作")
mute(s, alpha_id)
else:
print(f"[MAIN] {alpha_id} 检测结果: 合法")
print("\n" + "=" * 60)
print(f"[MAIN] 全部处理完成, 共处理 {count} 个alpha")
print("=" * 60)
if __name__ == "__main__": if __name__ == "__main__":

Write Preview
Loading…
Cancel
Save