AlgoRover Bittern Hour-Open Maker Scalper v1.0.0 Visual — Strategy by AlgoRover

By AlgoRover

Performance Metrics

Description

1. Strategy Logic / 策略基础逻辑I study Bittern as a specific mean-reversion hypothesis: can deviations from the current UTC hour’s opening price be traded often enough to pay for execution costs and occasional larger losses? The reference market is Binance ONDO/USDT perpetual, BINANCE:ONDOUSDT.P, on standard 3-minute candles. I do not interpret the hourly open as proven fair value; it is simply the anchor this strategy tests.我研究 Bittern,是想检验一个具体的均值回归假设:价格偏离当前 UTC 小时开盘价后,回归交易的次数和收益,能否覆盖执行成本及偶发的大额亏损?参考品种为币安 ONDO/USDT 永续合约,TradingView 代码 BINANCE:ONDOUSDT.P,使用标准 3 分钟 K 线。我没有把小时开盘价当成已经被证明的公允价值,它只是这套策略接受历史检验的参照点。After the first 3-minute candle of an hour completes, its open becomes the anchor. While flat, the strategy quotes a bid 1.2% below it and an ask 1.2% above it, keeping each quote at least one tick away from the completed close. An entry removes the opposing quote. The exit target follows later hourly anchors but stays at least 0.1% beyond entry in the profitable direction. A separate stop-market is 6% from entry, and a market exit applies after 120 completed bars. To me, the essential trade-off is frequent small reversions against less frequent trend losses; “maker scalper” does not mean every exit is maker or every loss is small.每个小时第一根 3 分钟 K 线收盘后,以其开盘价更新锚点;空仓时在锚点下方 1.2% 挂买单、上方 1.2% 挂卖单,并使报价至少离已完成 K 线收盘价一个最小价格单位。一侧成交后撤销另一侧。退出目标随新小时锚点调整,但始终位于开仓价盈利方向至少 0.1% 处;另外设置距开仓价 6% 的止损市价单,持仓满 120 根已完成 K 线则市价退出。我关注的是频繁的小幅回归收益与较少发生的趋势亏损之间的交换关系,maker 剥头皮并不意味着所有退出都是 maker,也不意味着每次亏损都很小。Figure 1. Rules against actual candles / 图1:规则与实际K线对照TradingView screenshot, BINANCE:ONDOUSDT.P, 3-minute candles, 12–13 November 2026; chart labels UTC+8. Captured 16 November 2026.The orange line is the confirmed UTC hourly anchor, teal/purple lines are working entry quotes, green is the active target, and red is the active stop. In the exported example, 12 November 22:21 UTC+8 enters long at 0.3439 and 22:36 exits at 0.3481 (TP_MAKER): 0.3481 × (1 − 1.2%) = 0.3439228, rounded to the 0.3439 bid. The displayed lines follow completed-bar updates; they do not imply an order already existed at the opening of the first candle. The bottom panel reports the entire loaded research window, not just this enlarged example.橙线为完成首根K线后确认的 UTC 小时开盘锚点,青色/紫色线为实际工作中的入场报价,绿色为持仓目标,红色为持仓止损。导出实例中,11 月 12 日 22:21(UTC+8)以 0.3439 做多,22:36 以 0.3481 退出(TP_MAKER):0.3481 ×(1 − 1.2%)= 0.3439228,按价格精度取为 0.3439。线条随完成 K 线更新,并不表示首根 K 线开盘时订单就已存在。下方面板统计整个已载入研究区间,不是仅统计本图放大的案例。2. Parameters and Rules / 参数规则I retain the supplied defaults: band 1.2%, minimum target distance 0.1%, stop 6%, maximum hold 120 bars (approximately six hours), both directions enabled, one-tick requoting, and a 3% quantity-resize threshold. Initial capital is 100,000 USDT; each opening target is 95% of current equity at 1x, with a 0.1 ONDO quantity step. The native exposure envelope is 2.27 times equity and the risk cap is 3 times, counting both resting opening orders; these are limits, not instructions to take a 3x position. Two resting quotes can require substantially more reserved margin than the eventual position.我保留压缩包默认规则:挂单带宽 1.2%、最小目标距离 0.1%、止损 6%、最长持仓 120 根 K 线(约六小时)、允许多空、报价变化至少一个 tick 才改单,目标数量变化超过 3% 时撤单并在后续完成 K 线重挂。初始资金为 100,000 USDT,按当前权益的 95%、1 倍杠杆确定每侧目标数量,ONDO 数量步长为 0.1。原生策略的总敞口范围为权益的 2.27 倍,风险上限为 3 倍,计算时包含两侧待成交开仓单;这些是限制,不代表实际开出 3 倍仓位。两侧挂单需要的预留保证金也可能明显高于最终持仓。The requested fees are Maker 0.01% and Taker 0.025%, per fill. Pine’s Strategy Tester uses one uniform 0.01% fee; the script separately deducts another 0.015% for stop, time, terminal, margin-call and marketable-target exits, and uses the adjusted equity for subsequent sizing. A resting target is classified as maker; a target repriced through the completed close is classified as taker. This is an order-intent proxy, not exchange liquidity verification. The adjusted curve and table therefore matter more than the Tester headline. Changing the Tester commission without changing the accounting invalidates this fee reconciliation.本次按每次成交收取 maker 0.01%、taker 0.025%。Pine 测试器先统一扣除 0.01%,脚本再对止损、超时、期末、保证金强平及穿越已完成收盘价的目标退出补扣 0.015%,后续定量使用补扣后的权益。已静置的目标单按 maker 归类,重新报价时穿越已完成收盘价的目标单按 taker 归类。这是根据订单行为估算流动性类别,并非交易所逐笔确认,因此应以调整后的曲线和面板理解分费率结果;直接修改测试器佣金而不同步调整核算,会破坏当前费用对账。I treat the Pine port as a visual research approximation. The Python source uses two independent post-only opening orders and cancellation after a fill; Pine uses emulator OCA cancellation and cannot reproduce cancel races, partial-fill sequencing, real queue priority, venue rejection/restart safeguards or reserved-order margin. The 1x TradingView margin model also generates partial liquidations. I have kept these differences visible rather than changing leverage or order size to improve the displayed return.我把这份 Pine 版本定位为可视化研究近似实现。Python 使用两张独立 post-only 开仓单,并在成交后撤销另一张;Pine 使用模拟器 OCA 撤单,不能复现撤单竞态、部分成交顺序、真实排队优先级、交易所拒单与重启保护,以及待成交订单的保证金预留。TradingView 的 1 倍保证金模型还会产生部分强平。我保留并披露这些差异,没有为了改善展示收益而调整杠杆或下单比例。3. Backtest Results / 回测效果I use the retained AlgoRover native run from 15 November 2025 to 15 November 2026 (UTC), whose source file matches this supplied package byte for byte and whose actual fill fees were rechecked. Starting from 100,000 USDT, it finished at 175,030.83 USDT: net profit 75,030.83 USDT, return and exact-365-day annualized return 75.03%, maximum equity drawdown 42.49%, 2,443 completed position lifecycles, win rate 89.36%, and approximately 200.79 completed trades per 30 days on average. Daily log-return Sharpe annualized with 365 days is 1.192; the native report’s separate 252-day convention reports 0.992. The retained 90-day run returned 35.23%, with 18.56% maximum drawdown and 632 completed lifecycles. Neither run simulates funding.我采用已归档的 AlgoRover 原生回测,区间为 2025 年 11 月 15 日至 2026 年 11 月 15 日(UTC);其运行时源码与本次压缩包逐字节相同,本次也重新核对了实际成交费率。100,000 USDT 初始资金最终为 175,030.83 USDT,净利润 75,030.83 USDT,总收益与完整 365 天年化收益均为 75.03%,最大权益回撤 42.49%,完整持仓生命周期 2,443 笔,胜率 89.36%,区间平均约 200.79 笔/30 天。按日对数收益、365 天年化计算的夏普为 1.192;原生报告另用 252 天口径,显示 0.992。已归档的 90 天结果为收益 35.23%、最大回撤 18.56%、完整交易 632 笔。两组结果均未模拟资金费。The verified TradingView run is much shorter: loaded chart history starts on 26 October 2026, and the script stops trading at 15 November 2026 00:00 UTC. It reports +2,300.68 USDT before the additional taker adjustment, approximately +1,634.26 USDT (+1.63%) after it, and 13.22% maximum drawdown sampled on completed-bar passes. The Tester’s own intrabar maximum drawdown is 13.74%, with 146 closed trade records including 25 margin-call records; grouping split exits gives 121 distinct entry lifecycles. These are not a full-year replication. All five display switches were tested together: the entire 292-row trade CSV was unchanged when the visuals were disabled.本次实际验证的 TradingView 区间明显更短:载入历史从 2026 年 10 月 26 日开始,脚本在 2026 年 11 月 15 日 00:00 UTC 结束交易。测试器在额外 taker 补扣前显示 +2,300.68 USDT,补扣后约 +1,634.26 USDT(+1.63%),按完成 K 线计算通过时点采样的最大回撤为 13.22%。测试器自身的盘中最大回撤为 13.74%,146 条平仓记录中包含 25 条保证金强平记录,将拆分退出按入场归并后为 121 个入场生命周期。这不是对全年结果的复现。五个显示开关已一起验证:全部关闭前后,完整 292 行成交 CSV 完全一致。My interpretation is that the native strategy is profitable in this historical sample but carries substantial path risk. An 89% win rate alongside a 42% drawdown is precisely why I would not summarize it as a stable high-win-rate system. Fees total 59,120.72 USDT in the annual run, and the short side loses about 5,015.88 USDT despite the positive aggregate result. Before treating this as executable performance, I would require conservative queue/slippage assumptions, funding, and a closer match of venue margin and fill sequencing. The current upload demonstrates the rules and their limits; it does not establish live reproducibility.我的判断是:原生策略在这段历史中盈利,但持有过程中的风险相当突出。约 89% 的胜率伴随约 42% 的回撤,正是我不会将其概括为稳定高胜率系统的原因。全年手续费累计 59,120.72 USDT,而且在总体盈利的同时,空头侧净亏损约 5,015.88 USDT。要进一步把它视为可以执行的业绩依据,我会要求更保守的排队和滑点假设、加入资金费,并让保证金机制和成交顺序更接近真实交易所。当前上传展示的是规则及其限制,尚不能证明实盘可以复现。Figure 2. Native equity and drawdown / 图2:原生净值与回撤Reconstructed from native timeseries.json, run 20261115-112033-414401; no TradingView results are substituted.I read both panels together: the final gain follows a long period below earlier equity peaks. The largest drawdown is 42.49% on 23 May 2026 at 20:00 UTC. This is peak-to-trough account-equity drawdown; it is not the 6% per-position stop setting, and it does not describe the maximum possible future loss.我会把上下两幅图一起看:最终收益之前,账户曾长时间低于此前权益高点。最大回撤 42.49% 出现在 2026 年 5 月 23 日 20:00 UTC。这是账户权益相对历史峰值的回撤,不是单笔持仓设置的 6% 止损,也不代表未来最大损失上限。This strategy was generated using AlgoRover and adapted into Pine Script for visual analysis and backtesting in TradingView, demonstrating a research workflow that moves from a trading idea to explicit rules, then to understanding, modification, and validation. AlgoRover supports more than 4,000 strategies across different categories, custom strategy generation, modification of existing strategy parameters and logic, extensive market data, and execution on local and cloud servers. Its positioning spans strategy generation, rule comprehension, logic modification, and historical validation, lowering the barriers to quantitative strategy research and implementation and making quantitative trading simpler. See the profile for details, and copy the command to install AlgoRover (codex can be replaced with claude or cursor):codex plugin marketplace add AlgoRover/algorover-hub --ref main该策略通过 AlgoRover 生成,并改写为Pine Script,以便在TradingView中进行图示观察与回测,展示了从交易想法到明确规则,再到理解、修改和验证的研究流程。AlgoRover支持 4,000余种不同类别的策略,支持策略自定义生成、现有策略参数与逻辑改写,并提供海量市场数据及本地、云端服务器运行能力。其定位贯穿策略生成、规则理解、逻辑改写和历史验证,降低量化策略研究与实现的门槛,让量化交易变得更简单。详情请看主页,直接复制命令安装AlgoRover:(codex可以替换成claude,cursor)codex plugin marketplace add AlgoRover/algorover-hub --ref main

Browse all 5,900+ TradingView Pine Script strategies

View on TradingView